CBCL~PM2.5*Age+Covariates Models with Model Checking
Claire Campbell; prior data cleaning by Hedyeh Ahmadi
Version October 23, 2023
R set up
Script not shown in the HTML file.
Note that the data cleaning and exploration for this analysis is in a separate file called written by Hedyeh Ahmadi.
## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.1 ✔ readr 2.1.4
## ✔ forcats 1.0.0 ✔ stringr 1.5.0
## ✔ ggplot2 3.4.2 ✔ tibble 3.2.1
## ✔ lubridate 1.9.2 ✔ tidyr 1.3.0
## ✔ purrr 1.0.1
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::group_rows() masks kableExtra::group_rows()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
## ------------------------------------------------------------------------------
##
## You have loaded plyr after dplyr - this is likely to cause problems.
## If you need functions from both plyr and dplyr, please load plyr first, then dplyr:
## library(plyr); library(dplyr)
##
## ------------------------------------------------------------------------------
##
##
## Attaching package: 'plyr'
##
##
## The following objects are masked from 'package:dplyr':
##
## arrange, count, desc, failwith, id, mutate, rename, summarise,
## summarize
##
##
## The following object is masked from 'package:purrr':
##
## compact
##
##
##
## Attaching package: 'scales'
##
##
## The following object is masked from 'package:purrr':
##
## discard
##
##
## The following object is masked from 'package:readr':
##
## col_factor
##
##
##
## Attaching package: 'ggpubr'
##
##
## The following object is masked from 'package:plyr':
##
## mutate
##
##
##
## Attaching package: 'matrixStats'
##
##
## The following object is masked from 'package:plyr':
##
## count
##
##
## The following object is masked from 'package:dplyr':
##
## count
##
##
## Loading required package: Matrix
##
##
## Attaching package: 'Matrix'
##
##
## The following objects are masked from 'package:tidyr':
##
## expand, pack, unpack
##
##
##
## Attaching package: 'nlme'
##
##
## The following object is masked from 'package:lme4':
##
## lmList
##
##
## The following object is masked from 'package:dplyr':
##
## collapse
##
##
##
## Attaching package: 'lmerTest'
##
##
## The following object is masked from 'package:lme4':
##
## lmer
##
##
## The following object is masked from 'package:stats':
##
## step
##
##
## Loading required package: HLMdiag
##
##
## Attaching package: 'HLMdiag'
##
##
## The following object is masked from 'package:stats':
##
## covratio
##
##
##
## Attaching package: 'Hmisc'
##
##
## The following object is masked from 'package:e1071':
##
## impute
##
##
## The following objects are masked from 'package:plyr':
##
## is.discrete, summarize
##
##
## The following objects are masked from 'package:dplyr':
##
## src, summarize
##
##
## The following objects are masked from 'package:xtable':
##
## label, label<-
##
##
## The following objects are masked from 'package:base':
##
## format.pval, units
##
##
##
## Attaching package: 'doBy'
##
##
## The following object is masked from 'package:dplyr':
##
## order_by
##
##
##
## Attaching package: 'rockchalk'
##
##
## The following object is masked from 'package:Hmisc':
##
## summarize
##
##
## The following objects are masked from 'package:e1071':
##
## kurtosis, skewness
##
##
## The following object is masked from 'package:plyr':
##
## summarize
##
##
## The following object is masked from 'package:dplyr':
##
## summarize
##
##
##
## Attaching package: 'MASS'
##
##
## The following object is masked from 'package:rockchalk':
##
## mvrnorm
##
##
## The following object is masked from 'package:dplyr':
##
## select
##
##
##
## Attaching package: 'psych'
##
##
## The following object is masked from 'package:Hmisc':
##
## describe
##
##
## The following objects are masked from 'package:scales':
##
## alpha, rescale
##
##
## The following objects are masked from 'package:ggplot2':
##
## %+%, alpha
##
##
## corrplot 0.92 loaded
##
## Loading required package: carData
##
##
## Attaching package: 'car'
##
##
## The following object is masked from 'package:psych':
##
## logit
##
##
## The following object is masked from 'package:dplyr':
##
## recode
##
##
## The following object is masked from 'package:purrr':
##
## some
##
##
## Loading required package: survival
##
##
## Attaching package: 'performance'
##
##
## The following object is masked from 'package:xtable':
##
## display
##
##
##
## Attaching package: 'arsenal'
##
##
## The following object is masked from 'package:Hmisc':
##
## %nin%
##
##
## The following object is masked from 'package:matrixStats':
##
## iqr
##
##
## The following object is masked from 'package:scales':
##
## ordinal
##
##
## The following object is masked from 'package:lubridate':
##
## is.Date
##
##
##
## Attaching package: 'NBZIMM'
##
##
## The following object is masked from 'package:psych':
##
## sim
##
##
## The following object is masked from 'package:stringr':
##
## fixed
##
##
##
## Attaching package: 'philentropy'
##
##
## The following objects are masked from 'package:psych':
##
## distance, manhattan, minkowskiImporting the Clean Data
HEI_Aim2_Long <- read.csv("HEI_Aim2_Long.csv")
HEI_Aim2_Wide <- read.csv("HEI_Aim2_Wide.csv")
HEI_Aim2_Long_1KidPerFamily <- read.csv("HEI_Aim2_Long_1KidPerFamily.csv")
dim(HEI_Aim2_Long)## [1] 30419 56dim(HEI_Aim2_Wide)## [1] 11866 140dim(HEI_Aim2_Long_1KidPerFamily)## [1] 25125 56names(HEI_Aim2_Long)## [1] "X" "subjectid"
## [3] "rel_family_id" "abcd_site"
## [5] "eventname" "rel_relationship"
## [7] "interview_date" "demo_l_p_select_language___1"
## [9] "cbcl_select_language___1" "rel_group_id"
## [11] "rel_ingroup_order" "rel_same_sex"
## [13] "reshist_addr1_pm252016aa" "sex"
## [15] "interview_age" "race_ethnicity"
## [17] "high.educ" "reshist_addr1_adi_perc"
## [19] "reshist_addr1_adi_wsum" "overall.income.b"
## [21] "overall.income.l" "overall.income.alltp"
## [23] "prnt.empl.bl" "prnt.empl.l"
## [25] "prnt.empl.alltp" "neighb_phenx_avg_p"
## [27] "neighb_phenx_sum_p" "reshist_addr1_popdensity"
## [29] "reshist_addr1_proxrd" "married"
## [31] "married.or.livingtogether" "cbcl_scr_syn_internal_r"
## [33] "cbcl_scr_syn_external_r" "cbcl_scr_syn_totprob_r"
## [35] "cbcl_scr_syn_anxdep_r" "cbcl_scr_syn_withdep_r"
## [37] "cbcl_scr_syn_attention_r" "cbcl_scr_syn_rulebreak_r"
## [39] "cbcl_scr_syn_aggressive_r" "cbcl_scr_syn_internal_t"
## [41] "cbcl_scr_syn_external_t" "cbcl_scr_syn_totprob_t"
## [43] "cbcl_scr_syn_anxdep_t" "cbcl_scr_syn_withdep_t"
## [45] "cbcl_scr_syn_attention_t" "cbcl_scr_syn_rulebreak_t"
## [47] "cbcl_scr_syn_aggressive_t" "reshist_addr1_years"
## [49] "income_midp" "demo_comb_income_v2"
## [51] "reshist_addr1_no2_2016_aavg" "reshist_addr1_o3_2016_annavg"
## [53] "reshist_addr1_pm252016aa_bl" "reshist_addr1_no2_2016_aavg_bl"
## [55] "reshist_addr1_o3_2016_annavg_bl" "high.educ_bl"names(HEI_Aim2_Wide)## [1] "X"
## [2] "subjectid"
## [3] "rel_family_id"
## [4] "rel_group_id"
## [5] "rel_same_sex"
## [6] "sex"
## [7] "race_ethnicity"
## [8] "reshist_addr1_pm252016aa.Baseline"
## [9] "abcd_site.Baseline"
## [10] "interview_date.Baseline"
## [11] "rel_relationship.Baseline"
## [12] "rel_ingroup_order.Baseline"
## [13] "high.educ.Baseline"
## [14] "interview_age.Baseline"
## [15] "demo_l_p_select_language___1.Baseline"
## [16] "cbcl_select_language___1.Baseline"
## [17] "reshist_addr1_adi_perc.Baseline"
## [18] "reshist_addr1_adi_wsum.Baseline"
## [19] "overall.income.b.Baseline"
## [20] "overall.income.l.Baseline"
## [21] "overall.income.alltp.Baseline"
## [22] "prnt.empl.bl.Baseline"
## [23] "prnt.empl.l.Baseline"
## [24] "prnt.empl.alltp.Baseline"
## [25] "neighb_phenx_avg_p.Baseline"
## [26] "neighb_phenx_sum_p.Baseline"
## [27] "reshist_addr1_popdensity.Baseline"
## [28] "reshist_addr1_proxrd.Baseline"
## [29] "married.Baseline"
## [30] "married.or.livingtogether.Baseline"
## [31] "cbcl_scr_syn_internal_r.Baseline"
## [32] "cbcl_scr_syn_external_r.Baseline"
## [33] "cbcl_scr_syn_totprob_r.Baseline"
## [34] "cbcl_scr_syn_anxdep_r.Baseline"
## [35] "cbcl_scr_syn_withdep_r.Baseline"
## [36] "cbcl_scr_syn_attention_r.Baseline"
## [37] "cbcl_scr_syn_rulebreak_r.Baseline"
## [38] "cbcl_scr_syn_aggressive_r.Baseline"
## [39] "cbcl_scr_syn_internal_t.Baseline"
## [40] "cbcl_scr_syn_external_t.Baseline"
## [41] "cbcl_scr_syn_totprob_t.Baseline"
## [42] "cbcl_scr_syn_anxdep_t.Baseline"
## [43] "cbcl_scr_syn_withdep_t.Baseline"
## [44] "cbcl_scr_syn_attention_t.Baseline"
## [45] "cbcl_scr_syn_rulebreak_t.Baseline"
## [46] "cbcl_scr_syn_aggressive_t.Baseline"
## [47] "reshist_addr1_years.Baseline"
## [48] "income_midp.Baseline"
## [49] "demo_comb_income_v2.Baseline"
## [50] "reshist_addr1_no2_2016_aavg.Baseline"
## [51] "reshist_addr1_o3_2016_annavg.Baseline"
## [52] "reshist_addr1_pm252016aa.1.year"
## [53] "abcd_site.1.year"
## [54] "interview_date.1.year"
## [55] "rel_relationship.1.year"
## [56] "rel_ingroup_order.1.year"
## [57] "high.educ.1.year"
## [58] "interview_age.1.year"
## [59] "demo_l_p_select_language___1.1.year"
## [60] "cbcl_select_language___1.1.year"
## [61] "reshist_addr1_adi_perc.1.year"
## [62] "reshist_addr1_adi_wsum.1.year"
## [63] "overall.income.b.1.year"
## [64] "overall.income.l.1.year"
## [65] "overall.income.alltp.1.year"
## [66] "prnt.empl.bl.1.year"
## [67] "prnt.empl.l.1.year"
## [68] "prnt.empl.alltp.1.year"
## [69] "neighb_phenx_avg_p.1.year"
## [70] "neighb_phenx_sum_p.1.year"
## [71] "reshist_addr1_popdensity.1.year"
## [72] "reshist_addr1_proxrd.1.year"
## [73] "married.1.year"
## [74] "married.or.livingtogether.1.year"
## [75] "cbcl_scr_syn_internal_r.1.year"
## [76] "cbcl_scr_syn_external_r.1.year"
## [77] "cbcl_scr_syn_totprob_r.1.year"
## [78] "cbcl_scr_syn_anxdep_r.1.year"
## [79] "cbcl_scr_syn_withdep_r.1.year"
## [80] "cbcl_scr_syn_attention_r.1.year"
## [81] "cbcl_scr_syn_rulebreak_r.1.year"
## [82] "cbcl_scr_syn_aggressive_r.1.year"
## [83] "cbcl_scr_syn_internal_t.1.year"
## [84] "cbcl_scr_syn_external_t.1.year"
## [85] "cbcl_scr_syn_totprob_t.1.year"
## [86] "cbcl_scr_syn_anxdep_t.1.year"
## [87] "cbcl_scr_syn_withdep_t.1.year"
## [88] "cbcl_scr_syn_attention_t.1.year"
## [89] "cbcl_scr_syn_rulebreak_t.1.year"
## [90] "cbcl_scr_syn_aggressive_t.1.year"
## [91] "reshist_addr1_years.1.year"
## [92] "income_midp.1.year"
## [93] "demo_comb_income_v2.1.year"
## [94] "reshist_addr1_no2_2016_aavg.1.year"
## [95] "reshist_addr1_o3_2016_annavg.1.year"
## [96] "reshist_addr1_pm252016aa.2.year"
## [97] "abcd_site.2.year"
## [98] "interview_date.2.year"
## [99] "rel_relationship.2.year"
## [100] "rel_ingroup_order.2.year"
## [101] "high.educ.2.year"
## [102] "interview_age.2.year"
## [103] "demo_l_p_select_language___1.2.year"
## [104] "cbcl_select_language___1.2.year"
## [105] "reshist_addr1_adi_perc.2.year"
## [106] "reshist_addr1_adi_wsum.2.year"
## [107] "overall.income.b.2.year"
## [108] "overall.income.l.2.year"
## [109] "overall.income.alltp.2.year"
## [110] "prnt.empl.bl.2.year"
## [111] "prnt.empl.l.2.year"
## [112] "prnt.empl.alltp.2.year"
## [113] "neighb_phenx_avg_p.2.year"
## [114] "neighb_phenx_sum_p.2.year"
## [115] "reshist_addr1_popdensity.2.year"
## [116] "reshist_addr1_proxrd.2.year"
## [117] "married.2.year"
## [118] "married.or.livingtogether.2.year"
## [119] "cbcl_scr_syn_internal_r.2.year"
## [120] "cbcl_scr_syn_external_r.2.year"
## [121] "cbcl_scr_syn_totprob_r.2.year"
## [122] "cbcl_scr_syn_anxdep_r.2.year"
## [123] "cbcl_scr_syn_withdep_r.2.year"
## [124] "cbcl_scr_syn_attention_r.2.year"
## [125] "cbcl_scr_syn_rulebreak_r.2.year"
## [126] "cbcl_scr_syn_aggressive_r.2.year"
## [127] "cbcl_scr_syn_internal_t.2.year"
## [128] "cbcl_scr_syn_external_t.2.year"
## [129] "cbcl_scr_syn_totprob_t.2.year"
## [130] "cbcl_scr_syn_anxdep_t.2.year"
## [131] "cbcl_scr_syn_withdep_t.2.year"
## [132] "cbcl_scr_syn_attention_t.2.year"
## [133] "cbcl_scr_syn_rulebreak_t.2.year"
## [134] "cbcl_scr_syn_aggressive_t.2.year"
## [135] "reshist_addr1_years.2.year"
## [136] "income_midp.2.year"
## [137] "demo_comb_income_v2.2.year"
## [138] "reshist_addr1_no2_2016_aavg.2.year"
## [139] "reshist_addr1_o3_2016_annavg.2.year"
## [140] "rel_relationship.1_year"names(HEI_Aim2_Long_1KidPerFamily)## [1] "X" "subjectid"
## [3] "rel_family_id" "abcd_site"
## [5] "eventname" "rel_relationship"
## [7] "interview_date" "demo_l_p_select_language___1"
## [9] "cbcl_select_language___1" "rel_group_id"
## [11] "rel_ingroup_order" "rel_same_sex"
## [13] "reshist_addr1_pm252016aa" "sex"
## [15] "interview_age" "race_ethnicity"
## [17] "high.educ" "reshist_addr1_adi_perc"
## [19] "reshist_addr1_adi_wsum" "overall.income.b"
## [21] "overall.income.l" "overall.income.alltp"
## [23] "prnt.empl.bl" "prnt.empl.l"
## [25] "prnt.empl.alltp" "neighb_phenx_avg_p"
## [27] "neighb_phenx_sum_p" "reshist_addr1_popdensity"
## [29] "reshist_addr1_proxrd" "married"
## [31] "married.or.livingtogether" "cbcl_scr_syn_internal_r"
## [33] "cbcl_scr_syn_external_r" "cbcl_scr_syn_totprob_r"
## [35] "cbcl_scr_syn_anxdep_r" "cbcl_scr_syn_withdep_r"
## [37] "cbcl_scr_syn_attention_r" "cbcl_scr_syn_rulebreak_r"
## [39] "cbcl_scr_syn_aggressive_r" "cbcl_scr_syn_internal_t"
## [41] "cbcl_scr_syn_external_t" "cbcl_scr_syn_totprob_t"
## [43] "cbcl_scr_syn_anxdep_t" "cbcl_scr_syn_withdep_t"
## [45] "cbcl_scr_syn_attention_t" "cbcl_scr_syn_rulebreak_t"
## [47] "cbcl_scr_syn_aggressive_t" "reshist_addr1_years"
## [49] "income_midp" "demo_comb_income_v2"
## [51] "reshist_addr1_no2_2016_aavg" "reshist_addr1_o3_2016_annavg"
## [53] "reshist_addr1_pm252016aa_bl" "reshist_addr1_no2_2016_aavg_bl"
## [55] "reshist_addr1_o3_2016_annavg_bl" "high.educ_bl"# Note we are keeping all families but choosing one kid per family
length(unique(HEI_Aim2_Long$rel_family_id))## [1] 9844length(unique(HEI_Aim2_Long$subjectid)) # matches number of rows of wide data :)## [1] 11866length(unique(HEI_Aim2_Long_1KidPerFamily$rel_family_id))## [1] 9844length(unique(HEI_Aim2_Long_1KidPerFamily$subjectid))## [1] 9844Copy baseline variables to all timepoints for full df
#rename so can use later
names(HEI_Aim2_Long)[names(HEI_Aim2_Long) == 'prnt.empl.bl'] <- 'prnt.empl.b'
#create dataset for table and comparison
baseline_vars <- subset(HEI_Aim2_Long, HEI_Aim2_Long$eventname=="Baseline", select = c("subjectid", "sex", "race_ethnicity", "high.educ", "neighb_phenx_avg_p", "overall.income.b", "prnt.empl.b"))
#rename variables
names(baseline_vars)[names(baseline_vars) == 'sex'] <- 'sex.bl'
names(baseline_vars)[names(baseline_vars) == 'race_ethnicity'] <- 'race_ethnicity.bl'
names(baseline_vars)[names(baseline_vars) == 'high.educ'] <- 'high.educ.bl'
names(baseline_vars)[names(baseline_vars) == 'neighb_phenx_avg_p'] <- 'neighb_phenx_avg_p.bl'
names(baseline_vars)[names(baseline_vars) == 'overall.income.b'] <- 'overall.income.bl'
names(baseline_vars)[names(baseline_vars) == 'prnt.empl.b'] <- 'prnt.empl.bl'
#add to initial df
HEI_Aim2_Long_2 <- merge(HEI_Aim2_Long, baseline_vars, by="subjectid")Descriptive Table before further Cleaning
#factor eventname
HEI_Aim2_Long_2$eventname <- as.factor(HEI_Aim2_Long_2$eventname)
HEI_Aim2_Long_2$eventname <- relevel(HEI_Aim2_Long_2$eventname , ref="Baseline")
#create smaller df
df_prior <- subset(HEI_Aim2_Long_2,select=c("subjectid","abcd_site","eventname","interview_age","reshist_addr1_pm252016aa_bl","prnt.empl.bl","overall.income.bl","sex.bl","race_ethnicity.bl","high.educ.bl","neighb_phenx_avg_p.bl","cbcl_scr_syn_internal_r","cbcl_scr_syn_external_r","cbcl_scr_syn_anxdep_r","cbcl_scr_syn_withdep_r","cbcl_scr_syn_attention_r","cbcl_scr_syn_rulebreak_r","cbcl_scr_syn_aggressive_r","cbcl_scr_syn_totprob_r"))
#create table
des_table_prior <- tableby(eventname ~ ., data = df_prior[ , -which(names(df_prior) %in% c("subjectid"))], total=F)
summary(des_table_prior, title = "Descriptive Statistics by Eventname Before Cleaning")##
##
## Table: Descriptive Statistics by Eventname Before Cleaning
##
## | | Baseline (N=11839) | 1-year (N=11200) | 2-year (N=7334) | p value|
## |:----------------------------------------|:------------------:|:-----------------:|:-----------------:|-------:|
## |**abcd_site** | | | | < 0.001|
## | site01 | 406 (3.4%) | 369 (3.3%) | 210 (2.9%) | |
## | site02 | 558 (4.7%) | 548 (4.9%) | 351 (4.8%) | |
## | site03 | 631 (5.3%) | 563 (5.0%) | 372 (5.1%) | |
## | site04 | 745 (6.3%) | 727 (6.5%) | 534 (7.3%) | |
## | site05 | 378 (3.2%) | 357 (3.2%) | 234 (3.2%) | |
## | site06 | 584 (4.9%) | 568 (5.1%) | 379 (5.2%) | |
## | site07 | 339 (2.9%) | 322 (2.9%) | 116 (1.6%) | |
## | site08 | 350 (3.0%) | 339 (3.0%) | 212 (2.9%) | |
## | site09 | 433 (3.7%) | 393 (3.5%) | 227 (3.1%) | |
## | site10 | 739 (6.2%) | 708 (6.3%) | 493 (6.7%) | |
## | site11 | 450 (3.8%) | 400 (3.6%) | 197 (2.7%) | |
## | site12 | 604 (5.1%) | 550 (4.9%) | 274 (3.7%) | |
## | site13 | 728 (6.1%) | 691 (6.2%) | 443 (6.0%) | |
## | site14 | 606 (5.1%) | 583 (5.2%) | 430 (5.9%) | |
## | site15 | 458 (3.9%) | 426 (3.8%) | 266 (3.6%) | |
## | site16 | 1011 (8.5%) | 979 (8.7%) | 640 (8.7%) | |
## | site17 | 578 (4.9%) | 562 (5.0%) | 374 (5.1%) | |
## | site18 | 384 (3.2%) | 376 (3.4%) | 223 (3.0%) | |
## | site19 | 550 (4.6%) | 521 (4.7%) | 397 (5.4%) | |
## | site20 | 707 (6.0%) | 687 (6.1%) | 528 (7.2%) | |
## | site21 | 600 (5.1%) | 531 (4.7%) | 434 (5.9%) | |
## |**interview_age** | | | | < 0.001|
## | Mean (SD) | 118.967 (7.495) | 131.073 (7.714) | 143.361 (7.747) | |
## | Range | 107.000 - 133.000 | 116.000 - 149.000 | 127.000 - 164.000 | |
## |**reshist_addr1_pm252016aa_bl** | | | | 0.745|
## | N-Miss | 651 | 587 | 224 | |
## | Mean (SD) | 7.663 (1.563) | 7.648 (1.561) | 7.650 (1.535) | |
## | Range | 1.722 - 15.902 | 1.722 - 15.902 | 1.722 - 15.902 | |
## |**prnt.empl.bl** | | | | 0.098|
## | N-Miss | 56 | 47 | 22 | |
## | Employed | 8194 (69.5%) | 7826 (70.2%) | 5214 (71.3%) | |
## | Other | 855 (7.3%) | 791 (7.1%) | 474 (6.5%) | |
## | Stay at Home Parent | 2065 (17.5%) | 1941 (17.4%) | 1262 (17.3%) | |
## | Unemployed | 669 (5.7%) | 595 (5.3%) | 362 (5.0%) | |
## |**overall.income.bl** | | | | 0.001|
## | N-Miss | 2 | 1 | 0 | |
## | [<50k] | 3215 (27.2%) | 2930 (26.2%) | 1833 (25.0%) | |
## | [>=100K] | 4544 (38.4%) | 4419 (39.5%) | 2952 (40.3%) | |
## | [>=50K & <100K] | 3065 (25.9%) | 2937 (26.2%) | 2000 (27.3%) | |
## | [Don't Know or Refuse] | 1013 (8.6%) | 913 (8.2%) | 549 (7.5%) | |
## |**sex.bl** | | | | 0.906|
## | Female | 5658 (47.8%) | 5335 (47.6%) | 3481 (47.5%) | |
## | Male | 6181 (52.2%) | 5865 (52.4%) | 3853 (52.5%) | |
## |**race_ethnicity.bl** | | | | < 0.001|
## | N-Miss | 2 | 2 | 0 | |
## | Asian | 250 (2.1%) | 239 (2.1%) | 158 (2.2%) | |
## | Black | 1777 (15.0%) | 1594 (14.2%) | 874 (11.9%) | |
## | Hispanic | 2405 (20.3%) | 2220 (19.8%) | 1411 (19.2%) | |
## | Other | 1243 (10.5%) | 1171 (10.5%) | 724 (9.9%) | |
## | White | 6162 (52.1%) | 5974 (53.3%) | 4167 (56.8%) | |
## |**high.educ.bl** | | | | < 0.001|
## | N-Miss | 14 | 12 | 10 | |
## | < HS Diploma | 592 (5.0%) | 526 (4.7%) | 306 (4.2%) | |
## | Bachelor | 3006 (25.4%) | 2889 (25.8%) | 2002 (27.3%) | |
## | HS Diploma/GED | 1129 (9.5%) | 1007 (9.0%) | 568 (7.8%) | |
## | Post Graduate Degree | 4025 (34.0%) | 3919 (35.0%) | 2611 (35.6%) | |
## | Some College | 3073 (26.0%) | 2847 (25.4%) | 1837 (25.1%) | |
## |**neighb_phenx_avg_p.bl** | | | | 0.003|
## | N-Miss | 8 | 5 | 3 | |
## | Mean (SD) | 3.890 (0.975) | 3.903 (0.969) | 3.938 (0.942) | |
## | Range | 1.000 - 5.000 | 1.000 - 5.000 | 1.000 - 5.000 | |
## |**cbcl_scr_syn_internal_r** | | | | 0.100|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 5.043 (5.522) | 5.108 (5.551) | 4.930 (5.614) | |
## | Range | 0.000 - 51.000 | 0.000 - 48.000 | 0.000 - 50.000 | |
## |**cbcl_scr_syn_external_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 4.455 (5.867) | 4.176 (5.656) | 3.918 (5.479) | |
## | Range | 0.000 - 49.000 | 0.000 - 47.000 | 0.000 - 46.000 | |
## |**cbcl_scr_syn_anxdep_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 2.516 (3.062) | 2.540 (3.072) | 2.322 (2.971) | |
## | Range | 0.000 - 26.000 | 0.000 - 22.000 | 0.000 - 22.000 | |
## |**cbcl_scr_syn_withdep_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 1.034 (1.709) | 1.116 (1.778) | 1.201 (1.901) | |
## | Range | 0.000 - 15.000 | 0.000 - 14.000 | 0.000 - 16.000 | |
## |**cbcl_scr_syn_attention_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 2.977 (3.495) | 2.858 (3.431) | 2.692 (3.298) | |
## | Range | 0.000 - 20.000 | 0.000 - 19.000 | 0.000 - 19.000 | |
## |**cbcl_scr_syn_rulebreak_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 1.192 (1.861) | 1.120 (1.822) | 1.057 (1.833) | |
## | Range | 0.000 - 20.000 | 0.000 - 20.000 | 0.000 - 23.000 | |
## |**cbcl_scr_syn_aggressive_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 3.262 (4.355) | 3.056 (4.185) | 2.861 (3.990) | |
## | Range | 0.000 - 36.000 | 0.000 - 33.000 | 0.000 - 32.000 | |
## |**cbcl_scr_syn_totprob_r** | | | | < 0.001|
## | N-Miss | 8 | 18 | 5 | |
## | Mean (SD) | 18.178 (17.968) | 17.520 (17.567) | 16.388 (17.001) | |
## | Range | 0.000 - 139.000 | 0.000 - 128.000 | 0.000 - 161.000 | |Creating variables for modeling and tables
The following variables are time-invariant, will use baseline covariates since PM2.5 collected at baseline: - reshist_addr1_pm252016aa_bl which is the Baseline PM2.5. - reshist_addr1_no2_2016_aavg_bl which is the Baseline NO2. - sex.bl - race_ethnicity.bl - high.educ.bl - prnt.empl.bl - neighb_phenx_avg_p.bl - overall.income.bl
The following variables are time-varying: - all CBCL outcomes - interview_age
Copy baseline variables to all timepoints for 1KidPerFamily
#rename so can use later
names(HEI_Aim2_Long_1KidPerFamily)[names(HEI_Aim2_Long_1KidPerFamily) == 'prnt.empl.bl'] <- 'prnt.empl.b'
#create dataset for table and comparison
baseline_vars_1KidPerFamily <- subset(HEI_Aim2_Long_1KidPerFamily, HEI_Aim2_Long_1KidPerFamily$eventname=="Baseline", select = c("subjectid", "sex", "race_ethnicity", "high.educ", "neighb_phenx_avg_p", "overall.income.b"))
#rename variables
names(baseline_vars_1KidPerFamily)[names(baseline_vars_1KidPerFamily) == 'sex'] <- 'sex.bl'
names(baseline_vars_1KidPerFamily)[names(baseline_vars_1KidPerFamily) == 'race_ethnicity'] <- 'race_ethnicity.bl'
names(baseline_vars_1KidPerFamily)[names(baseline_vars_1KidPerFamily) == 'high.educ'] <- 'high.educ.bl'
names(baseline_vars_1KidPerFamily)[names(baseline_vars_1KidPerFamily) == 'neighb_phenx_avg_p'] <- 'neighb_phenx_avg_p.bl'
names(baseline_vars_1KidPerFamily)[names(baseline_vars_1KidPerFamily) == 'overall.income.b'] <- 'overall.income.bl'
names(baseline_vars_1KidPerFamily)[names(baseline_vars_1KidPerFamily) == 'prnt.empl.b'] <- 'prnt.empl.bl'
#add to initial df
HEI_Aim2_Long_1KidPerFamily_2 <- merge(HEI_Aim2_Long_1KidPerFamily, baseline_vars, by="subjectid")Clean 1KidPerFamily
## Cleaning
#merge Asian into Other group b/c statistically Asian group is too small
tapply(HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl,
HEI_Aim2_Long_1KidPerFamily_2$eventname,table, useNA = "always")## $`1-year`
##
## Asian Black Hispanic Other White <NA>
## 217 1334 1932 964 4803 1
##
## $`2-year`
##
## Asian Black Hispanic Other White <NA>
## 141 715 1232 597 3326 0
##
## $Baseline
##
## Asian Black Hispanic Other White <NA>
## 228 1496 2101 1029 4963 1HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl <-
ifelse(HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl=="Asian","Other",
HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl)
tapply(HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl,
HEI_Aim2_Long_1KidPerFamily_2$eventname,table, useNA = "always")## $`1-year`
##
## Black Hispanic Other White <NA>
## 1334 1932 1181 4803 1
##
## $`2-year`
##
## Black Hispanic Other White <NA>
## 715 1232 738 3326 0
##
## $Baseline
##
## Black Hispanic Other White <NA>
## 1496 2101 1257 4963 1#reformat variables
HEI_Aim2_Long_1KidPerFamily_2$eventname <-
as.factor(HEI_Aim2_Long_1KidPerFamily_2$eventname)
HEI_Aim2_Long_1KidPerFamily_2$eventname <-
relevel(HEI_Aim2_Long_1KidPerFamily_2$eventname , ref="Baseline")
table(HEI_Aim2_Long_1KidPerFamily_2$eventname, useNA = "always")##
## Baseline 1-year 2-year <NA>
## 9818 9251 6011 0HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_internal_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_internal_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_external_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_external_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_anxdep_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_anxdep_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_withdep_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_withdep_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_attention_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_attention_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_rulebreak_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_rulebreak_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_aggressive_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_aggressive_r)
HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_totprob_r <- as.numeric(HEI_Aim2_Long_1KidPerFamily_2$cbcl_scr_syn_totprob_r)
HEI_Aim2_Long_1KidPerFamily_2$abcd_site <- as.factor(HEI_Aim2_Long_1KidPerFamily_2$abcd_site)
HEI_Aim2_Long_1KidPerFamily_2$subjectid <- as.factor(HEI_Aim2_Long_1KidPerFamily_2$subjectid)
HEI_Aim2_Long_1KidPerFamily_2$prnt.empl.bl <- factor(HEI_Aim2_Long_1KidPerFamily_2$prnt.empl.bl, levels = c("Employed", "Stay at Home Parent", "Unemployed", "Other"))
HEI_Aim2_Long_1KidPerFamily_2$overall.income.bl <- factor(HEI_Aim2_Long_1KidPerFamily_2$overall.income.bl, levels = c("[>=100K]", "[>=50K & <100K]", "[<50k]", "[Don't Know or Refuse]"))
HEI_Aim2_Long_1KidPerFamily_2$sex.bl <- factor(HEI_Aim2_Long_1KidPerFamily_2$sex.bl, levels = c("Male", "Female"))
HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl <- factor(HEI_Aim2_Long_1KidPerFamily_2$race_ethnicity.bl, levels = c("White", "Hispanic", "Black", "Other"))
HEI_Aim2_Long_1KidPerFamily_2$high.educ.bl <- factor(HEI_Aim2_Long_1KidPerFamily_2$high.educ.bl, levels = c("Post Graduate Degree", "Bachelor", "Some College", "HS Diploma/GED", "< HS Diploma"))Create Final Dataset
#create smaller df
df <- subset(HEI_Aim2_Long_1KidPerFamily_2,select=c("subjectid","abcd_site","eventname","interview_age","reshist_addr1_pm252016aa_bl","reshist_addr1_no2_2016_aavg_bl","prnt.empl.bl","overall.income.bl","sex.bl","race_ethnicity.bl","high.educ.bl","neighb_phenx_avg_p.bl","cbcl_scr_syn_internal_r","cbcl_scr_syn_external_r","cbcl_scr_syn_anxdep_r","cbcl_scr_syn_withdep_r","cbcl_scr_syn_attention_r","cbcl_scr_syn_rulebreak_r","cbcl_scr_syn_aggressive_r","cbcl_scr_syn_totprob_r"))
#complete cases because needed for zinb
df_cc <- df[complete.cases(df),]
#percentage of subjects lost with complete.cases
lost_sub <- data.frame(table(df$eventname))
colnames(lost_sub) <- c("eventname","abcd")
interim <- data.frame(table(df_cc$eventname))[2]
lost_sub$sample <- interim$Freq
lost_sub$diff <- lost_sub$abcd - lost_sub$sample
lost_sub$percent <- lost_sub$diff/lost_sub$abcd
#center age at 9years-old (i.e., 108 months)
df_cc$interview_age.c9 <- df_cc$interview_age-108
#change to years
df_cc$interview_age.c9.y <- df_cc$interview_age.c9/12
#center pm2.5 to 5 (recommended by WHO)
df_cc$reshist_addr1_pm252016aa_bl.c5 <- df_cc$reshist_addr1_pm252016aa_bl-5
#center no2 to 5.33 (recommended by WHO)
df_cc$reshist_addr1_no2_2016_aavg_bl.c533 <- df_cc$reshist_addr1_no2_2016_aavg_bl-5.33
#center around mean
neighb_phenx_avg_p.bl.cm <- df_cc$neighb_phenx_avg_p.bl - mean(df_cc$neighb_phenx_avg_p.bl)
#create table
des_table <- tableby(eventname ~ ., data = df_cc[ , -which(names(df_cc) %in% c("subjectid"))], total=F)
summary(des_table, title = "Descriptive Statistics by Eventname After Cleaning")| Baseline (N=9271) | 1-year (N=8759) | 2-year (N=5827) | p value | |
|---|---|---|---|---|
| abcd_site | < 0.001 | |||
| site01 | 323 (3.5%) | 299 (3.4%) | 184 (3.2%) | |
| site02 | 305 (3.3%) | 298 (3.4%) | 210 (3.6%) | |
| site03 | 556 (6.0%) | 493 (5.6%) | 330 (5.7%) | |
| site04 | 631 (6.8%) | 615 (7.0%) | 451 (7.7%) | |
| site05 | 322 (3.5%) | 305 (3.5%) | 203 (3.5%) | |
| site06 | 509 (5.5%) | 495 (5.7%) | 334 (5.7%) | |
| site07 | 290 (3.1%) | 275 (3.1%) | 102 (1.8%) | |
| site08 | 302 (3.3%) | 289 (3.3%) | 183 (3.1%) | |
| site09 | 404 (4.4%) | 366 (4.2%) | 214 (3.7%) | |
| site10 | 623 (6.7%) | 592 (6.8%) | 427 (7.3%) | |
| site11 | 382 (4.1%) | 339 (3.9%) | 166 (2.8%) | |
| site12 | 481 (5.2%) | 438 (5.0%) | 236 (4.1%) | |
| site13 | 617 (6.7%) | 585 (6.7%) | 387 (6.6%) | |
| site14 | 359 (3.9%) | 344 (3.9%) | 255 (4.4%) | |
| site15 | 398 (4.3%) | 371 (4.2%) | 232 (4.0%) | |
| site16 | 794 (8.6%) | 769 (8.8%) | 500 (8.6%) | |
| site17 | 509 (5.5%) | 495 (5.7%) | 336 (5.8%) | |
| site18 | 351 (3.8%) | 344 (3.9%) | 206 (3.5%) | |
| site19 | 216 (2.3%) | 208 (2.4%) | 185 (3.2%) | |
| site20 | 447 (4.8%) | 433 (4.9%) | 327 (5.6%) | |
| site21 | 452 (4.9%) | 406 (4.6%) | 359 (6.2%) | |
| interview_age | < 0.001 | |||
| Mean (SD) | 118.856 (7.411) | 130.924 (7.617) | 143.081 (7.635) | |
| Range | 107.000 - 133.000 | 117.000 - 149.000 | 127.000 - 164.000 | |
| reshist_addr1_pm252016aa_bl | 0.771 | |||
| Mean (SD) | 7.706 (1.571) | 7.693 (1.571) | 7.690 (1.552) | |
| Range | 1.722 - 15.902 | 1.722 - 15.902 | 1.722 - 15.902 | |
| reshist_addr1_no2_2016_aavg_bl | 0.972 | |||
| Mean (SD) | 18.595 (5.571) | 18.579 (5.567) | 18.575 (5.573) | |
| Range | 0.729 - 37.940 | 0.729 - 37.940 | 0.729 - 35.346 | |
| prnt.empl.bl | 0.373 | |||
| Employed | 6444 (69.5%) | 6146 (70.2%) | 4139 (71.0%) | |
| Stay at Home Parent | 1612 (17.4%) | 1516 (17.3%) | 1002 (17.2%) | |
| Unemployed | 539 (5.8%) | 481 (5.5%) | 299 (5.1%) | |
| Other | 676 (7.3%) | 616 (7.0%) | 387 (6.6%) | |
| overall.income.bl | 0.018 | |||
| [>=100K] | 3514 (37.9%) | 3418 (39.0%) | 2307 (39.6%) | |
| [>=50K & <100K] | 2419 (26.1%) | 2312 (26.4%) | 1597 (27.4%) | |
| [<50k] | 2564 (27.7%) | 2335 (26.7%) | 1494 (25.6%) | |
| [Don’t Know or Refuse] | 774 (8.3%) | 694 (7.9%) | 429 (7.4%) | |
| sex.bl | 0.861 | |||
| Male | 4858 (52.4%) | 4605 (52.6%) | 3080 (52.9%) | |
| Female | 4413 (47.6%) | 4154 (47.4%) | 2747 (47.1%) | |
| race_ethnicity.bl | < 0.001 | |||
| White | 4759 (51.3%) | 4612 (52.7%) | 3236 (55.5%) | |
| Hispanic | 1958 (21.1%) | 1800 (20.6%) | 1198 (20.6%) | |
| Black | 1363 (14.7%) | 1221 (13.9%) | 678 (11.6%) | |
| Other | 1191 (12.8%) | 1126 (12.9%) | 715 (12.3%) | |
| high.educ.bl | 0.002 | |||
| Post Graduate Degree | 3179 (34.3%) | 3090 (35.3%) | 2093 (35.9%) | |
| Bachelor | 2310 (24.9%) | 2217 (25.3%) | 1551 (26.6%) | |
| Some College | 2423 (26.1%) | 2247 (25.7%) | 1466 (25.2%) | |
| HS Diploma/GED | 899 (9.7%) | 800 (9.1%) | 460 (7.9%) | |
| < HS Diploma | 460 (5.0%) | 405 (4.6%) | 257 (4.4%) | |
| neighb_phenx_avg_p.bl | 0.034 | |||
| Mean (SD) | 3.873 (0.976) | 3.884 (0.971) | 3.915 (0.948) | |
| Range | 1.000 - 5.000 | 1.000 - 5.000 | 1.000 - 5.000 | |
| cbcl_scr_syn_internal_r | 0.042 | |||
| Mean (SD) | 5.154 (5.539) | 5.281 (5.612) | 5.047 (5.610) | |
| Range | 0.000 - 51.000 | 0.000 - 48.000 | 0.000 - 50.000 | |
| cbcl_scr_syn_external_r | < 0.001 | |||
| Mean (SD) | 4.484 (5.798) | 4.232 (5.622) | 3.949 (5.368) | |
| Range | 0.000 - 49.000 | 0.000 - 46.000 | 0.000 - 46.000 | |
| cbcl_scr_syn_anxdep_r | < 0.001 | |||
| Mean (SD) | 2.569 (3.060) | 2.613 (3.091) | 2.352 (2.941) | |
| Range | 0.000 - 26.000 | 0.000 - 22.000 | 0.000 - 22.000 | |
| cbcl_scr_syn_withdep_r | < 0.001 | |||
| Mean (SD) | 1.045 (1.700) | 1.151 (1.800) | 1.235 (1.912) | |
| Range | 0.000 - 14.000 | 0.000 - 14.000 | 0.000 - 16.000 | |
| cbcl_scr_syn_attention_r | < 0.001 | |||
| Mean (SD) | 3.042 (3.508) | 2.941 (3.458) | 2.782 (3.324) | |
| Range | 0.000 - 19.000 | 0.000 - 19.000 | 0.000 - 19.000 | |
| cbcl_scr_syn_rulebreak_r | < 0.001 | |||
| Mean (SD) | 1.204 (1.844) | 1.142 (1.823) | 1.068 (1.829) | |
| Range | 0.000 - 20.000 | 0.000 - 20.000 | 0.000 - 23.000 | |
| cbcl_scr_syn_aggressive_r | < 0.001 | |||
| Mean (SD) | 3.280 (4.305) | 3.090 (4.154) | 2.881 (3.894) | |
| Range | 0.000 - 36.000 | 0.000 - 33.000 | 0.000 - 32.000 | |
| cbcl_scr_syn_totprob_r | < 0.001 | |||
| Mean (SD) | 18.519 (17.931) | 17.986 (17.659) | 16.755 (16.864) | |
| Range | 0.000 - 139.000 | 0.000 - 128.000 | 0.000 - 161.000 | |
| interview_age.c9 | < 0.001 | |||
| Mean (SD) | 10.856 (7.411) | 22.924 (7.617) | 35.081 (7.635) | |
| Range | -1.000 - 25.000 | 9.000 - 41.000 | 19.000 - 56.000 | |
| interview_age.c9.y | < 0.001 | |||
| Mean (SD) | 0.905 (0.618) | 1.910 (0.635) | 2.923 (0.636) | |
| Range | -0.083 - 2.083 | 0.750 - 3.417 | 1.583 - 4.667 | |
| reshist_addr1_pm252016aa_bl.c5 | 0.771 | |||
| Mean (SD) | 2.706 (1.571) | 2.693 (1.571) | 2.690 (1.552) | |
| Range | -3.278 - 10.902 | -3.278 - 10.902 | -3.278 - 10.902 | |
| reshist_addr1_no2_2016_aavg_bl.c533 | 0.972 | |||
| Mean (SD) | 13.265 (5.571) | 13.249 (5.567) | 13.245 (5.573) | |
| Range | -4.601 - 32.610 | -4.601 - 32.610 | -4.601 - 30.016 |
Final_DF_Descriptives <- summary(des_table, title = "Descriptive Statistics by Eventname After Cleaning")
#write.csv(Final_DF_Descriptives, "Descriptives_Final_Dataset.csv")Internalizing
CBCL + AP Longitudinal Models
Zero-Inflated (ZI) Negative Binomial (NB): glmm.zinb in NBZIMM package.
CBCL only needs one nested random intercept since we eliminated the family nesting by choosing one kid per family.
For the negative binomial portion of the model, we do not nest by subject since the ICC across subjects is very low.
internal_zinb_r <- glmm.zinb(cbcl_scr_syn_internal_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 7
## Computational time: 1.206 minutessummary(internal_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1084049
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 0.8512357 1.107149
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_internal_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) 1.3153073 0.05105041 14510
## reshist_addr1_pm252016aa_bl.c5 0.0078568 0.01111178 9307
## interview_age.c9.y 0.0275201 0.00878765 14510
## race_ethnicity.blHispanic -0.0243546 0.03135380 9307
## race_ethnicity.blBlack -0.3586827 0.03532015 9307
## race_ethnicity.blOther -0.0507550 0.03170926 9307
## high.educ.blBachelor 0.0198012 0.02650623 9307
## high.educ.blSome College 0.0489022 0.03038442 9307
## high.educ.blHS Diploma/GED -0.1305187 0.04313049 9307
## high.educ.bl< HS Diploma -0.1089449 0.05555887 9307
## prnt.empl.blStay at Home Parent 0.0306533 0.02695206 9307
## prnt.empl.blUnemployed 0.1290158 0.04441220 9307
## prnt.empl.blOther 0.2001879 0.03882581 9307
## neighb_phenx_avg_p.bl.cm -0.1216138 0.01132191 9307
## overall.income.bl[>=50K & <100K] 0.1087366 0.02683571 9307
## overall.income.bl[<50k] 0.1703172 0.03382476 9307
## overall.income.bl[Don't Know or Refuse] 0.0708793 0.04256789 9307
## sex.blFemale 0.0500218 0.01969583 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0060889 0.00262873 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0092719 0.00281105 14510
## t-value p-value
## (Intercept) 25.764874 0.0000
## reshist_addr1_pm252016aa_bl.c5 0.707073 0.4795
## interview_age.c9.y 3.131677 0.0017
## race_ethnicity.blHispanic -0.776768 0.4373
## race_ethnicity.blBlack -10.155185 0.0000
## race_ethnicity.blOther -1.600636 0.1095
## high.educ.blBachelor 0.747041 0.4551
## high.educ.blSome College 1.609448 0.1076
## high.educ.blHS Diploma/GED -3.026135 0.0025
## high.educ.bl< HS Diploma -1.960891 0.0499
## prnt.empl.blStay at Home Parent 1.137326 0.2554
## prnt.empl.blUnemployed 2.904963 0.0037
## prnt.empl.blOther 5.156053 0.0000
## neighb_phenx_avg_p.bl.cm -10.741460 0.0000
## overall.income.bl[>=50K & <100K] 4.051937 0.0001
## overall.income.bl[<50k] 5.035282 0.0000
## overall.income.bl[Don't Know or Refuse] 1.665088 0.0959
## sex.blFemale 2.539716 0.0111
## reshist_addr1_no2_2016_aavg_bl.c533 -2.316304 0.0206
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -3.298374 0.0010
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.398
## interview_age.c9.y -0.304 0.375
## race_ethnicity.blHispanic -0.027 -0.072 0.004
## race_ethnicity.blBlack -0.022 -0.025 0.007
## race_ethnicity.blOther -0.085 -0.020 0.004
## high.educ.blBachelor -0.185 0.000 0.000
## high.educ.blSome College -0.126 -0.014 0.003
## high.educ.blHS Diploma/GED -0.075 -0.007 0.005
## high.educ.bl< HS Diploma -0.028 -0.016 -0.002
## prnt.empl.blStay at Home Parent -0.084 -0.016 0.002
## prnt.empl.blUnemployed -0.026 -0.003 0.000
## prnt.empl.blOther -0.039 0.003 0.007
## neighb_phenx_avg_p.bl.cm -0.182 0.052 -0.003
## overall.income.bl[>=50K & <100K] -0.124 -0.016 0.001
## overall.income.bl[<50k] -0.061 -0.027 -0.001
## overall.income.bl[Don't Know or Refuse] -0.059 -0.027 -0.002
## sex.blFemale -0.186 -0.003 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 -0.519 -0.233 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.262 -0.432 -0.867
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.349
## race_ethnicity.blOther 0.286 0.256
## high.educ.blBachelor -0.019 -0.013 -0.001
## high.educ.blSome College -0.111 -0.084 -0.024 0.455
## high.educ.blHS Diploma/GED -0.143 -0.142 -0.006 0.330
## high.educ.bl< HS Diploma -0.166 -0.071 -0.009 0.262
## prnt.empl.blStay at Home Parent 0.044 0.091 0.018 -0.028
## prnt.empl.blUnemployed 0.010 -0.040 0.010 -0.009
## prnt.empl.blOther 0.040 0.009 -0.012 -0.013
## neighb_phenx_avg_p.bl.cm 0.029 0.137 0.040 -0.004
## overall.income.bl[>=50K & <100K] -0.091 -0.060 -0.014 -0.174
## overall.income.bl[<50k] -0.145 -0.181 -0.081 -0.159
## overall.income.bl[Don't Know or Refuse] -0.097 -0.124 -0.061 -0.100
## sex.blFemale -0.007 -0.017 -0.018 0.015
## reshist_addr1_no2_2016_aavg_bl.c533 -0.055 -0.082 -0.030 0.014
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.002 -0.002 -0.001
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.494
## high.educ.bl< HS Diploma 0.406 0.372
## prnt.empl.blStay at Home Parent -0.014 -0.050 -0.094
## prnt.empl.blUnemployed -0.009 -0.068 -0.097 0.147
## prnt.empl.blOther -0.033 -0.010 -0.019 0.158
## neighb_phenx_avg_p.bl.cm 0.061 0.056 0.051 0.027
## overall.income.bl[>=50K & <100K] -0.277 -0.171 -0.113 -0.031
## overall.income.bl[<50k] -0.417 -0.364 -0.307 -0.054
## overall.income.bl[Don't Know or Refuse] -0.250 -0.237 -0.218 -0.077
## sex.blFemale 0.023 0.015 -0.004 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 0.021 0.005 -0.018 0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.004 0.004
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.131
## neighb_phenx_avg_p.bl.cm 0.021 0.004
## overall.income.bl[>=50K & <100K] -0.015 -0.050 0.081
## overall.income.bl[<50k] -0.101 -0.139 0.150 0.505
## overall.income.bl[Don't Know or Refuse] -0.077 -0.098 0.083 0.359
## sex.blFemale 0.020 0.020 0.026 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.010 -0.003 0.103 -0.010
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 -0.006 0.001 -0.002
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.482
## sex.blFemale -0.007 0.008
## reshist_addr1_no2_2016_aavg_bl.c533 -0.018 -0.004 0.000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.002 -0.001 -0.005
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -3.0452108 -0.7882604 -0.1957453 0.4336318 4.2772592
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(internal_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.3899559 0.6076061
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -3.926007 0.16280472 23817
## reshist_addr1_pm252016aa_bl.c5 -0.028105 0.03615469 23817
## interview_age.c9.y -0.049121 0.04707073 23817
## race_ethnicity.blHispanic 0.066927 0.07726332 23817
## race_ethnicity.blBlack 0.893290 0.07379195 23817
## race_ethnicity.blOther 0.385189 0.07534511 23817
## high.educ.blBachelor 0.130870 0.06638952 23817
## high.educ.blSome College 0.051135 0.07583844 23817
## high.educ.blHS Diploma/GED 0.595885 0.09101122 23817
## high.educ.bl< HS Diploma 0.813104 0.10627461 23817
## prnt.empl.blStay at Home Parent -0.051963 0.06520361 23817
## prnt.empl.blUnemployed -0.119281 0.09325067 23817
## prnt.empl.blOther 0.005704 0.08731933 23817
## neighb_phenx_avg_p.bl.cm 0.254243 0.02675268 23817
## overall.income.bl[>=50K & <100K] -0.072396 0.06919594 23817
## overall.income.bl[<50k] 0.012813 0.08024803 23817
## overall.income.bl[Don't Know or Refuse] 0.295554 0.08931692 23817
## sex.blFemale -0.075385 0.04597174 23817
## reshist_addr1_no2_2016_aavg_bl.c533 -0.003520 0.00668834 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.059262 0.01403646 23817
## t-value p-value
## (Intercept) -24.114823 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.777346 0.4370
## interview_age.c9.y -1.043557 0.2967
## race_ethnicity.blHispanic 0.866221 0.3864
## race_ethnicity.blBlack 12.105525 0.0000
## race_ethnicity.blOther 5.112332 0.0000
## high.educ.blBachelor 1.971244 0.0487
## high.educ.blSome College 0.674256 0.5002
## high.educ.blHS Diploma/GED 6.547383 0.0000
## high.educ.bl< HS Diploma 7.650974 0.0000
## prnt.empl.blStay at Home Parent -0.796939 0.4255
## prnt.empl.blUnemployed -1.279149 0.2009
## prnt.empl.blOther 0.065321 0.9479
## neighb_phenx_avg_p.bl.cm 9.503460 0.0000
## overall.income.bl[>=50K & <100K] -1.046247 0.2955
## overall.income.bl[<50k] 0.159666 0.8731
## overall.income.bl[Don't Know or Refuse] 3.309046 0.0009
## sex.blFemale -1.639817 0.1011
## reshist_addr1_no2_2016_aavg_bl.c533 -0.526272 0.5987
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 4.221971 0.0000
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.512
## interview_age.c9.y -0.532 0.628
## race_ethnicity.blHispanic -0.031 -0.051 0.002
## race_ethnicity.blBlack -0.044 -0.014 0.021
## race_ethnicity.blOther -0.096 0.001 0.009
## high.educ.blBachelor -0.156 0.006 -0.004
## high.educ.blSome College -0.105 -0.005 0.007
## high.educ.blHS Diploma/GED -0.081 0.010 0.018
## high.educ.bl< HS Diploma -0.033 -0.010 -0.001
## prnt.empl.blStay at Home Parent -0.065 -0.019 -0.004
## prnt.empl.blUnemployed -0.027 0.007 0.009
## prnt.empl.blOther -0.035 0.013 0.020
## neighb_phenx_avg_p.bl.cm -0.151 0.032 -0.002
## overall.income.bl[>=50K & <100K] -0.091 -0.009 0.002
## overall.income.bl[<50k] -0.045 -0.019 -0.007
## overall.income.bl[Don't Know or Refuse] -0.050 -0.029 0.000
## sex.blFemale -0.132 -0.004 0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.413 -0.168 -0.003
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.469 -0.738 -0.875
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.464
## race_ethnicity.blOther 0.353 0.342
## high.educ.blBachelor -0.013 -0.006 0.007
## high.educ.blSome College -0.114 -0.104 -0.009 0.479
## high.educ.blHS Diploma/GED -0.159 -0.160 0.002 0.410
## high.educ.bl< HS Diploma -0.186 -0.103 -0.003 0.357
## prnt.empl.blStay at Home Parent 0.046 0.106 0.023 -0.023
## prnt.empl.blUnemployed 0.017 -0.021 0.013 -0.018
## prnt.empl.blOther 0.057 0.012 -0.008 -0.017
## neighb_phenx_avg_p.bl.cm 0.017 0.130 0.032 0.004
## overall.income.bl[>=50K & <100K] -0.105 -0.090 -0.024 -0.166
## overall.income.bl[<50k] -0.150 -0.206 -0.078 -0.158
## overall.income.bl[Don't Know or Refuse] -0.117 -0.152 -0.068 -0.118
## sex.blFemale -0.003 -0.026 -0.017 0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.066 -0.103 -0.044 0.004
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.003 -0.007 -0.008 0.000
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.588
## high.educ.bl< HS Diploma 0.524 0.542
## prnt.empl.blStay at Home Parent -0.013 -0.057 -0.125
## prnt.empl.blUnemployed -0.005 -0.082 -0.121 0.176
## prnt.empl.blOther -0.031 -0.022 -0.036 0.165
## neighb_phenx_avg_p.bl.cm 0.049 0.050 0.053 0.033
## overall.income.bl[>=50K & <100K] -0.283 -0.207 -0.155 -0.025
## overall.income.bl[<50k] -0.409 -0.412 -0.386 -0.051
## overall.income.bl[Don't Know or Refuse] -0.295 -0.308 -0.285 -0.094
## sex.blFemale 0.009 0.011 -0.014 0.012
## reshist_addr1_no2_2016_aavg_bl.c533 0.010 -0.010 -0.034 0.009
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.003 -0.006 0.007 0.016
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.149
## neighb_phenx_avg_p.bl.cm 0.021 -0.010
## overall.income.bl[>=50K & <100K] -0.014 -0.048 0.061
## overall.income.bl[<50k] -0.081 -0.125 0.125 0.536
## overall.income.bl[Don't Know or Refuse] -0.081 -0.109 0.071 0.438
## sex.blFemale 0.036 0.019 0.019 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.011 -0.007 0.097 -0.019
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.006 -0.018 0.001 -0.003
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.603
## sex.blFemale -0.004 -0.005
## reshist_addr1_no2_2016_aavg_bl.c533 -0.031 -0.006 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.006 0.006 0.002 0.004
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -0.8729333 -0.3076123 -0.2379597 -0.1768534 18.1306493
##
## Number of Observations: 23857
## Number of Groups: 21anova(internal_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 2669.5251 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 0.0222 0.8815
## interview_age.c9.y 1 14510 0.2566 0.6125
## race_ethnicity.bl 3 9307 19.4934 <.0001
## high.educ.bl 4 9307 12.1081 <.0001
## prnt.empl.bl 3 9307 15.3185 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 132.8621 <.0001
## overall.income.bl 3 9307 9.8506 <.0001
## sex.bl 1 9307 6.4381 0.0112
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 5.4363 0.0197
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 10.8793 0.0010VarCorr(internal_zinb_r)## Variance StdDev
## abcd_site = pdLogChol(1)
## (Intercept) 0.01175163 0.1084049
## subjectid = pdLogChol(1)
## (Intercept) 0.72460220 0.8512357
## Residual 1.22577990 1.1071494Assumption checking for ZINB Models
Zero inflated negative binomial (zinb) regression already has overdispersion and excess zeros and this is accounted for in the zinb modeling chosen, “The data distribution combines the negative binomial distribution and the logit distribution”
Details on zinb can be found here: link
For Model Checking we will follow the following pdf: link This info is further detailed/published in books by Cameron and Trivedi (2013) and Hilbe (2014) and in Garay, Hashimoto, Ortega, and Lachos (2011).
They suggest using Pearson residuals.
#Check outlier/residuals with this df
internal_res <- df_cc
internal_res$level1_resid.raw <- residuals(internal_zinb_r)
internal_res$level1_resid.pearson <- residuals(internal_zinb_r, type="pearson")
#Add predicted values (Yhat)
internal_res$cbcl_scr_syn_internal_r_predicted <- predict(internal_zinb_r,internal_res,type="response")
#Incidence
internal_res$incidence <- estimate.probability(internal_res$cbcl_scr_syn_internal_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(internal_res$level1_resid.pearson)
Incidence vs. X’s Plots
“These plots show each of the independent variables plotted against the incidence as measured by Y (CBCL Outcome). They should be scanned for outliers and curvilinear patterns.”
#age
ggplot(internal_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(internal_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Residuals vs Y (CBCL Outcome) Plot
“This plot shows the residuals versus the dependent variable. It can be used to spot outliers.”
plot(internal_res$level1_resid.pearson, internal_res$cbcl_scr_syn_internal_r)
Residuals vs Yhat Plot
“This plot shows the residuals versus the predicted value (Yhat) of the dependent variable. It can show outliers.”
plot(internal_res$level1_resid.pearson, internal_res$cbcl_scr_syn_internal_r_predicted)
Residuals vs Row Plot
“This plot shows the residuals versus the row numbers. It is used to quickly spot rows that have large residuals.”
plot(as.numeric(rownames(internal_res)),internal_res$level1_resid.pearson)
Residuals vs X’s Plots
“These plots show the residuals plotted against the independent variables. They are used to spot outliers. They are also used to find curvilinear patterns that are not represented in the regression model.”
#age
ggplot(internal_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(internal_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
For below models, view Internalizing above for notes.
Externalizing
CBCL + AP Longitudinal Models
external_zinb_r <- glmm.zinb(cbcl_scr_syn_external_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 9
## Computational time: 1.443 minutessummary(external_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1294082
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 1.099431 1.057932
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_external_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) 1.0211060 0.06320481 14510
## reshist_addr1_pm252016aa_bl.c5 -0.0077650 0.01363091 9307
## interview_age.c9.y -0.0239486 0.00957918 14510
## race_ethnicity.blHispanic -0.0547048 0.04004575 9307
## race_ethnicity.blBlack -0.1035225 0.04439903 9307
## race_ethnicity.blOther -0.0375336 0.04074581 9307
## high.educ.blBachelor 0.1079517 0.03414719 9307
## high.educ.blSome College 0.1944569 0.03891004 9307
## high.educ.blHS Diploma/GED 0.0686692 0.05456791 9307
## high.educ.bl< HS Diploma 0.1027116 0.07027984 9307
## prnt.empl.blStay at Home Parent 0.0120180 0.03457321 9307
## prnt.empl.blUnemployed 0.1996353 0.05599056 9307
## prnt.empl.blOther 0.1979445 0.04940589 9307
## neighb_phenx_avg_p.bl.cm -0.1291003 0.01441675 9307
## overall.income.bl[>=50K & <100K] 0.1279780 0.03450922 9307
## overall.income.bl[<50k] 0.2649776 0.04315572 9307
## overall.income.bl[Don't Know or Refuse] 0.1667453 0.05412054 9307
## sex.blFemale -0.2976364 0.02524457 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0049374 0.00332575 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0054209 0.00302668 14510
## t-value p-value
## (Intercept) 16.155511 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.569658 0.5689
## interview_age.c9.y -2.500071 0.0124
## race_ethnicity.blHispanic -1.366059 0.1720
## race_ethnicity.blBlack -2.331639 0.0197
## race_ethnicity.blOther -0.921165 0.3570
## high.educ.blBachelor 3.161363 0.0016
## high.educ.blSome College 4.997603 0.0000
## high.educ.blHS Diploma/GED 1.258417 0.2083
## high.educ.bl< HS Diploma 1.461466 0.1439
## prnt.empl.blStay at Home Parent 0.347609 0.7281
## prnt.empl.blUnemployed 3.565517 0.0004
## prnt.empl.blOther 4.006495 0.0001
## neighb_phenx_avg_p.bl.cm -8.954881 0.0000
## overall.income.bl[>=50K & <100K] 3.708516 0.0002
## overall.income.bl[<50k] 6.140035 0.0000
## overall.income.bl[Don't Know or Refuse] 3.080999 0.0021
## sex.blFemale -11.790112 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -1.484605 0.1377
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -1.791028 0.0733
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.385
## interview_age.c9.y -0.262 0.325
## race_ethnicity.blHispanic -0.024 -0.078 0.003
## race_ethnicity.blBlack -0.021 -0.030 0.006
## race_ethnicity.blOther -0.086 -0.025 0.003
## high.educ.blBachelor -0.194 0.001 0.000
## high.educ.blSome College -0.133 -0.015 0.002
## high.educ.blHS Diploma/GED -0.081 -0.008 0.005
## high.educ.bl< HS Diploma -0.032 -0.016 -0.001
## prnt.empl.blStay at Home Parent -0.085 -0.017 0.002
## prnt.empl.blUnemployed -0.026 -0.003 0.000
## prnt.empl.blOther -0.039 0.002 0.007
## neighb_phenx_avg_p.bl.cm -0.186 0.056 -0.003
## overall.income.bl[>=50K & <100K] -0.131 -0.016 0.000
## overall.income.bl[<50k] -0.067 -0.028 -0.001
## overall.income.bl[Don't Know or Refuse] -0.064 -0.029 -0.002
## sex.blFemale -0.186 -0.004 0.003
## reshist_addr1_no2_2016_aavg_bl.c533 -0.534 -0.235 0.003
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.226 -0.373 -0.870
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.357
## race_ethnicity.blOther 0.288 0.263
## high.educ.blBachelor -0.022 -0.016 -0.002
## high.educ.blSome College -0.111 -0.085 -0.025 0.461
## high.educ.blHS Diploma/GED -0.144 -0.147 -0.009 0.339
## high.educ.bl< HS Diploma -0.168 -0.077 -0.012 0.268
## prnt.empl.blStay at Home Parent 0.043 0.092 0.017 -0.030
## prnt.empl.blUnemployed 0.010 -0.040 0.010 -0.009
## prnt.empl.blOther 0.041 0.011 -0.011 -0.014
## neighb_phenx_avg_p.bl.cm 0.030 0.136 0.042 -0.004
## overall.income.bl[>=50K & <100K] -0.088 -0.061 -0.011 -0.175
## overall.income.bl[<50k] -0.143 -0.182 -0.079 -0.160
## overall.income.bl[Don't Know or Refuse] -0.096 -0.125 -0.056 -0.100
## sex.blFemale -0.008 -0.019 -0.017 0.013
## reshist_addr1_no2_2016_aavg_bl.c533 -0.059 -0.083 -0.031 0.014
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.001 -0.001 -0.001
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.503
## high.educ.bl< HS Diploma 0.413 0.381
## prnt.empl.blStay at Home Parent -0.015 -0.050 -0.095
## prnt.empl.blUnemployed -0.010 -0.069 -0.098 0.149
## prnt.empl.blOther -0.033 -0.014 -0.020 0.159
## neighb_phenx_avg_p.bl.cm 0.061 0.055 0.049 0.028
## overall.income.bl[>=50K & <100K] -0.276 -0.173 -0.115 -0.029
## overall.income.bl[<50k] -0.417 -0.367 -0.310 -0.052
## overall.income.bl[Don't Know or Refuse] -0.253 -0.241 -0.220 -0.075
## sex.blFemale 0.022 0.014 -0.005 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 0.022 0.008 -0.016 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.002 0.003
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.134
## neighb_phenx_avg_p.bl.cm 0.023 0.003
## overall.income.bl[>=50K & <100K] -0.014 -0.049 0.080
## overall.income.bl[<50k] -0.101 -0.139 0.151 0.509
## overall.income.bl[Don't Know or Refuse] -0.079 -0.099 0.083 0.364
## sex.blFemale 0.018 0.017 0.027 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.011 -0.002 0.099 -0.009
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 -0.006 0.001 -0.001
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.490
## sex.blFemale -0.007 0.007
## reshist_addr1_no2_2016_aavg_bl.c533 -0.016 -0.003 0.000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.002 0.000 -0.003
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -3.2018245 -0.7290504 -0.2623209 0.3851928 4.4818057
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(external_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.2681476 0.5423461
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -4.121234 0.12997565 23817
## reshist_addr1_pm252016aa_bl.c5 0.055019 0.03041678 23817
## interview_age.c9.y 0.200675 0.03769513 23817
## race_ethnicity.blHispanic 0.203178 0.06360254 23817
## race_ethnicity.blBlack 0.594394 0.06584425 23817
## race_ethnicity.blOther 0.366721 0.06103961 23817
## high.educ.blBachelor 0.106629 0.05358877 23817
## high.educ.blSome College 0.143541 0.06233709 23817
## high.educ.blHS Diploma/GED 0.414071 0.08155326 23817
## high.educ.bl< HS Diploma 0.549831 0.09859802 23817
## prnt.empl.blStay at Home Parent 0.018552 0.05396768 23817
## prnt.empl.blUnemployed 0.022966 0.08288402 23817
## prnt.empl.blOther -0.295705 0.08714322 23817
## neighb_phenx_avg_p.bl.cm 0.203195 0.02329844 23817
## overall.income.bl[>=50K & <100K] -0.203062 0.05597617 23817
## overall.income.bl[<50k] -0.171169 0.06783270 23817
## overall.income.bl[Don't Know or Refuse] 0.165372 0.07685796 23817
## sex.blFemale 0.222933 0.03896640 23817
## reshist_addr1_no2_2016_aavg_bl.c533 -0.008428 0.00542719 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.004830 0.01158174 23817
## t-value p-value
## (Intercept) -31.70774 0.0000
## reshist_addr1_pm252016aa_bl.c5 1.80884 0.0705
## interview_age.c9.y 5.32364 0.0000
## race_ethnicity.blHispanic 3.19449 0.0014
## race_ethnicity.blBlack 9.02727 0.0000
## race_ethnicity.blOther 6.00791 0.0000
## high.educ.blBachelor 1.98977 0.0466
## high.educ.blSome College 2.30266 0.0213
## high.educ.blHS Diploma/GED 5.07731 0.0000
## high.educ.bl< HS Diploma 5.57649 0.0000
## prnt.empl.blStay at Home Parent 0.34376 0.7310
## prnt.empl.blUnemployed 0.27709 0.7817
## prnt.empl.blOther -3.39333 0.0007
## neighb_phenx_avg_p.bl.cm 8.72139 0.0000
## overall.income.bl[>=50K & <100K] -3.62766 0.0003
## overall.income.bl[<50k] -2.52340 0.0116
## overall.income.bl[Don't Know or Refuse] 2.15165 0.0314
## sex.blFemale 5.72117 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -1.55300 0.1204
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.41706 0.6766
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.543
## interview_age.c9.y -0.577 0.639
## race_ethnicity.blHispanic -0.038 -0.046 0.007
## race_ethnicity.blBlack -0.043 -0.009 0.027
## race_ethnicity.blOther -0.096 -0.003 0.010
## high.educ.blBachelor -0.157 0.001 -0.006
## high.educ.blSome College -0.109 -0.009 0.002
## high.educ.blHS Diploma/GED -0.076 0.004 0.014
## high.educ.bl< HS Diploma -0.025 -0.026 -0.008
## prnt.empl.blStay at Home Parent -0.073 -0.015 0.001
## prnt.empl.blUnemployed -0.025 0.002 0.005
## prnt.empl.blOther -0.030 0.012 0.016
## neighb_phenx_avg_p.bl.cm -0.167 0.036 0.000
## overall.income.bl[>=50K & <100K] -0.087 -0.006 0.001
## overall.income.bl[<50k] -0.037 -0.022 -0.007
## overall.income.bl[Don't Know or Refuse] -0.043 -0.031 -0.003
## sex.blFemale -0.165 -0.005 0.007
## reshist_addr1_no2_2016_aavg_bl.c533 -0.422 -0.163 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.493 -0.746 -0.859
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.407
## race_ethnicity.blOther 0.334 0.305
## high.educ.blBachelor -0.008 -0.005 0.012
## high.educ.blSome College -0.116 -0.092 -0.009 0.459
## high.educ.blHS Diploma/GED -0.154 -0.151 0.002 0.362
## high.educ.bl< HS Diploma -0.172 -0.087 0.001 0.305
## prnt.empl.blStay at Home Parent 0.046 0.104 0.021 -0.031
## prnt.empl.blUnemployed 0.014 -0.031 0.013 -0.016
## prnt.empl.blOther 0.041 0.002 -0.013 -0.018
## neighb_phenx_avg_p.bl.cm 0.022 0.128 0.031 0.001
## overall.income.bl[>=50K & <100K] -0.097 -0.080 -0.020 -0.159
## overall.income.bl[<50k] -0.141 -0.199 -0.079 -0.150
## overall.income.bl[Don't Know or Refuse] -0.107 -0.140 -0.064 -0.108
## sex.blFemale -0.003 -0.018 -0.012 0.012
## reshist_addr1_no2_2016_aavg_bl.c533 -0.064 -0.098 -0.039 0.011
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.012 -0.008 0.003
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.538
## high.educ.bl< HS Diploma 0.469 0.464
## prnt.empl.blStay at Home Parent -0.019 -0.058 -0.116
## prnt.empl.blUnemployed -0.009 -0.078 -0.116 0.164
## prnt.empl.blOther -0.027 -0.011 -0.027 0.140
## neighb_phenx_avg_p.bl.cm 0.054 0.051 0.055 0.030
## overall.income.bl[>=50K & <100K] -0.285 -0.189 -0.135 -0.023
## overall.income.bl[<50k] -0.413 -0.396 -0.355 -0.048
## overall.income.bl[Don't Know or Refuse] -0.284 -0.289 -0.270 -0.091
## sex.blFemale 0.017 0.012 -0.004 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 0.018 -0.002 -0.021 0.011
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.003 0.016 0.012
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.124
## neighb_phenx_avg_p.bl.cm 0.026 -0.005
## overall.income.bl[>=50K & <100K] -0.012 -0.041 0.073
## overall.income.bl[<50k] -0.088 -0.117 0.133 0.490
## overall.income.bl[Don't Know or Refuse] -0.083 -0.098 0.085 0.389
## sex.blFemale 0.031 0.013 0.023 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.013 -0.006 0.110 -0.013
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.014 0.000 -0.002
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.547
## sex.blFemale -0.005 0.000
## reshist_addr1_no2_2016_aavg_bl.c533 -0.022 0.000 0.007
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.008 0.009 0.001 -0.006
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -0.8295843 -0.3553204 -0.2848921 0.2312446 17.3194845
##
## Number of Observations: 23857
## Number of Groups: 21anova(external_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 918.4594 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 0.5341 0.4649
## interview_age.c9.y 1 14510 69.9272 <.0001
## race_ethnicity.bl 3 9307 5.9760 0.0005
## high.educ.bl 4 9307 31.8373 <.0001
## prnt.empl.bl 3 9307 15.2776 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 91.7969 <.0001
## overall.income.bl 3 9307 12.2969 <.0001
## sex.bl 1 9307 139.0157 <.0001
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 2.2187 0.1364
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 3.2078 0.0733VarCorr(external_zinb_r)## Variance StdDev
## abcd_site = pdLogChol(1)
## (Intercept) 0.01674648 0.1294082
## subjectid = pdLogChol(1)
## (Intercept) 1.20874949 1.0994314
## Residual 1.11922050 1.0579322Assumption checking for ZINB Models
#Check outlier/residuals with this df
external_res <- df_cc
external_res$level1_resid.raw <- residuals(external_zinb_r)
external_res$level1_resid.pearson <- residuals(external_zinb_r, type="pearson")
#Add predicted values (Yhat)
external_res$cbcl_scr_syn_external_r_predicted <- predict(external_zinb_r,external_res,type="response")
#Incidence
external_res$incidence <- estimate.probability(external_res$cbcl_scr_syn_external_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(external_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(external_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(external_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
### Residuals vs Y (CBCL Outcome) Plot
plot(external_res$level1_resid.pearson, external_res$cbcl_scr_syn_external_r)
### Residuals vs Yhat Plot
plot(external_res$level1_resid.pearson, external_res$cbcl_scr_syn_external_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(external_res)),external_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(external_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(external_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Anxious/Depressed
CBCL + AP Longitudinal Models
anxdep_zinb_r <- glmm.zinb(cbcl_scr_syn_anxdep_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 11
## Computational time: 1.845 minutessummary(anxdep_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1232991
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 0.971808 0.9387779
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_anxdep_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) 0.6762767 0.05866669 14510
## reshist_addr1_pm252016aa_bl.c5 -0.0028472 0.01283374 9307
## interview_age.c9.y 0.0005011 0.01010981 14510
## race_ethnicity.blHispanic -0.0124020 0.03628362 9307
## race_ethnicity.blBlack -0.4364621 0.04131043 9307
## race_ethnicity.blOther -0.0965066 0.03679521 9307
## high.educ.blBachelor -0.0283915 0.03065143 9307
## high.educ.blSome College -0.0539359 0.03524092 9307
## high.educ.blHS Diploma/GED -0.2846316 0.05035520 9307
## high.educ.bl< HS Diploma -0.2716848 0.06496712 9307
## prnt.empl.blStay at Home Parent 0.0397649 0.03125715 9307
## prnt.empl.blUnemployed 0.1843156 0.05153048 9307
## prnt.empl.blOther 0.1504990 0.04525515 9307
## neighb_phenx_avg_p.bl.cm -0.1134722 0.01316139 9307
## overall.income.bl[>=50K & <100K] 0.1123188 0.03104744 9307
## overall.income.bl[<50k] 0.1616208 0.03927431 9307
## overall.income.bl[Don't Know or Refuse] 0.0351743 0.04956511 9307
## sex.blFemale 0.0551860 0.02285260 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0063207 0.00303574 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0093103 0.00327250 14510
## t-value p-value
## (Intercept) 11.527440 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.221853 0.8244
## interview_age.c9.y 0.049564 0.9605
## race_ethnicity.blHispanic -0.341808 0.7325
## race_ethnicity.blBlack -10.565423 0.0000
## race_ethnicity.blOther -2.622802 0.0087
## high.educ.blBachelor -0.926271 0.3543
## high.educ.blSome College -1.530489 0.1259
## high.educ.blHS Diploma/GED -5.652478 0.0000
## high.educ.bl< HS Diploma -4.181881 0.0000
## prnt.empl.blStay at Home Parent 1.272186 0.2033
## prnt.empl.blUnemployed 3.576828 0.0003
## prnt.empl.blOther 3.325566 0.0009
## neighb_phenx_avg_p.bl.cm -8.621592 0.0000
## overall.income.bl[>=50K & <100K] 3.617652 0.0003
## overall.income.bl[<50k] 4.115177 0.0000
## overall.income.bl[Don't Know or Refuse] 0.709658 0.4779
## sex.blFemale 2.414869 0.0158
## reshist_addr1_no2_2016_aavg_bl.c533 -2.082111 0.0374
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -2.845008 0.0044
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.397
## interview_age.c9.y -0.301 0.373
## race_ethnicity.blHispanic -0.027 -0.073 0.004
## race_ethnicity.blBlack -0.023 -0.025 0.006
## race_ethnicity.blOther -0.085 -0.021 0.003
## high.educ.blBachelor -0.184 -0.001 0.000
## high.educ.blSome College -0.125 -0.015 0.003
## high.educ.blHS Diploma/GED -0.073 -0.009 0.005
## high.educ.bl< HS Diploma -0.027 -0.017 -0.002
## prnt.empl.blStay at Home Parent -0.084 -0.016 0.002
## prnt.empl.blUnemployed -0.026 -0.003 0.000
## prnt.empl.blOther -0.039 0.002 0.007
## neighb_phenx_avg_p.bl.cm -0.184 0.054 -0.003
## overall.income.bl[>=50K & <100K] -0.126 -0.015 0.001
## overall.income.bl[<50k] -0.062 -0.026 -0.001
## overall.income.bl[Don't Know or Refuse] -0.059 -0.027 -0.002
## sex.blFemale -0.189 -0.002 0.003
## reshist_addr1_no2_2016_aavg_bl.c533 -0.521 -0.234 0.004
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.258 -0.430 -0.865
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.344
## race_ethnicity.blOther 0.285 0.253
## high.educ.blBachelor -0.018 -0.012 0.000
## high.educ.blSome College -0.109 -0.082 -0.024 0.453
## high.educ.blHS Diploma/GED -0.141 -0.141 -0.006 0.327
## high.educ.bl< HS Diploma -0.166 -0.068 -0.009 0.258
## prnt.empl.blStay at Home Parent 0.043 0.092 0.020 -0.031
## prnt.empl.blUnemployed 0.010 -0.040 0.010 -0.009
## prnt.empl.blOther 0.040 0.010 -0.011 -0.013
## neighb_phenx_avg_p.bl.cm 0.029 0.139 0.039 -0.005
## overall.income.bl[>=50K & <100K] -0.092 -0.059 -0.015 -0.175
## overall.income.bl[<50k] -0.146 -0.179 -0.081 -0.160
## overall.income.bl[Don't Know or Refuse] -0.096 -0.120 -0.061 -0.100
## sex.blFemale -0.008 -0.018 -0.018 0.015
## reshist_addr1_no2_2016_aavg_bl.c533 -0.054 -0.081 -0.029 0.014
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 0.000 -0.002 -0.001
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.488
## high.educ.bl< HS Diploma 0.401 0.367
## prnt.empl.blStay at Home Parent -0.015 -0.053 -0.094
## prnt.empl.blUnemployed -0.009 -0.069 -0.097 0.147
## prnt.empl.blOther -0.033 -0.011 -0.020 0.157
## neighb_phenx_avg_p.bl.cm 0.062 0.057 0.051 0.027
## overall.income.bl[>=50K & <100K] -0.277 -0.169 -0.111 -0.031
## overall.income.bl[<50k] -0.418 -0.362 -0.306 -0.052
## overall.income.bl[Don't Know or Refuse] -0.249 -0.233 -0.215 -0.076
## sex.blFemale 0.023 0.015 -0.003 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 0.021 0.006 -0.018 0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.004 0.004
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.131
## neighb_phenx_avg_p.bl.cm 0.022 0.004
## overall.income.bl[>=50K & <100K] -0.014 -0.049 0.082
## overall.income.bl[<50k] -0.102 -0.140 0.149 0.503
## overall.income.bl[Don't Know or Refuse] -0.078 -0.097 0.083 0.355
## sex.blFemale 0.020 0.020 0.027 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.011 -0.003 0.102 -0.009
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.003 -0.006 0.001 -0.001
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.477
## sex.blFemale -0.007 0.008
## reshist_addr1_no2_2016_aavg_bl.c533 -0.017 -0.004 0.001
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 0.002 -0.001 -0.005
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.8149443 -0.7601946 -0.2212773 0.4359539 4.4148170
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(anxdep_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.4147199 0.4309516
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -4.909827 0.15368787 23817
## reshist_addr1_pm252016aa_bl.c5 -0.099370 0.03191419 23817
## interview_age.c9.y 0.043321 0.03961885 23817
## race_ethnicity.blHispanic 0.009406 0.06449590 23817
## race_ethnicity.blBlack 1.027369 0.05900214 23817
## race_ethnicity.blOther 0.315586 0.06450021 23817
## high.educ.blBachelor 0.355437 0.05622283 23817
## high.educ.blSome College 0.302606 0.06278469 23817
## high.educ.blHS Diploma/GED 0.498365 0.07692219 23817
## high.educ.bl< HS Diploma 0.896391 0.08538109 23817
## prnt.empl.blStay at Home Parent 0.007080 0.05330213 23817
## prnt.empl.blUnemployed -0.086601 0.07195668 23817
## prnt.empl.blOther 0.013143 0.06849761 23817
## neighb_phenx_avg_p.bl.cm 0.310656 0.02163439 23817
## overall.income.bl[>=50K & <100K] -0.206034 0.06018983 23817
## overall.income.bl[<50k] 0.286803 0.06437230 23817
## overall.income.bl[Don't Know or Refuse] 0.339290 0.07406045 23817
## sex.blFemale -0.326601 0.03809398 23817
## reshist_addr1_no2_2016_aavg_bl.c533 0.022451 0.00587493 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.101816 0.01151053 23817
## t-value p-value
## (Intercept) -31.94675 0.0000
## reshist_addr1_pm252016aa_bl.c5 -3.11367 0.0018
## interview_age.c9.y 1.09344 0.2742
## race_ethnicity.blHispanic 0.14584 0.8840
## race_ethnicity.blBlack 17.41240 0.0000
## race_ethnicity.blOther 4.89279 0.0000
## high.educ.blBachelor 6.32194 0.0000
## high.educ.blSome College 4.81973 0.0000
## high.educ.blHS Diploma/GED 6.47882 0.0000
## high.educ.bl< HS Diploma 10.49871 0.0000
## prnt.empl.blStay at Home Parent 0.13282 0.8943
## prnt.empl.blUnemployed -1.20351 0.2288
## prnt.empl.blOther 0.19187 0.8478
## neighb_phenx_avg_p.bl.cm 14.35935 0.0000
## overall.income.bl[>=50K & <100K] -3.42306 0.0006
## overall.income.bl[<50k] 4.45538 0.0000
## overall.income.bl[Don't Know or Refuse] 4.58126 0.0000
## sex.blFemale -8.57355 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 3.82154 0.0001
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 8.84544 0.0000
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.487
## interview_age.c9.y -0.520 0.654
## race_ethnicity.blHispanic -0.029 -0.043 -0.001
## race_ethnicity.blBlack -0.040 -0.011 0.021
## race_ethnicity.blOther -0.083 0.006 0.006
## high.educ.blBachelor -0.162 0.004 -0.003
## high.educ.blSome College -0.113 -0.007 0.009
## high.educ.blHS Diploma/GED -0.090 0.011 0.019
## high.educ.bl< HS Diploma -0.049 -0.009 0.001
## prnt.empl.blStay at Home Parent -0.064 -0.015 0.000
## prnt.empl.blUnemployed -0.029 0.010 0.014
## prnt.empl.blOther -0.032 0.014 0.019
## neighb_phenx_avg_p.bl.cm -0.130 0.022 -0.004
## overall.income.bl[>=50K & <100K] -0.081 -0.004 0.002
## overall.income.bl[<50k] -0.047 -0.017 -0.005
## overall.income.bl[Don't Know or Refuse] -0.051 -0.025 0.003
## sex.blFemale -0.103 -0.001 0.007
## reshist_addr1_no2_2016_aavg_bl.c533 -0.393 -0.169 -0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.464 -0.771 -0.882
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.493
## race_ethnicity.blOther 0.359 0.367
## high.educ.blBachelor -0.017 -0.003 0.006
## high.educ.blSome College -0.107 -0.099 -0.011 0.542
## high.educ.blHS Diploma/GED -0.143 -0.148 0.001 0.452
## high.educ.bl< HS Diploma -0.174 -0.104 -0.010 0.411
## prnt.empl.blStay at Home Parent 0.051 0.108 0.032 -0.017
## prnt.empl.blUnemployed 0.014 -0.013 0.018 -0.020
## prnt.empl.blOther 0.060 0.011 -0.001 -0.014
## neighb_phenx_avg_p.bl.cm 0.014 0.139 0.032 0.007
## overall.income.bl[>=50K & <100K] -0.103 -0.096 -0.024 -0.159
## overall.income.bl[<50k] -0.161 -0.212 -0.081 -0.165
## overall.income.bl[Don't Know or Refuse] -0.119 -0.151 -0.066 -0.120
## sex.blFemale -0.004 -0.036 -0.021 0.009
## reshist_addr1_no2_2016_aavg_bl.c533 -0.067 -0.115 -0.053 0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.007 0.000 -0.005 0.002
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.622
## high.educ.bl< HS Diploma 0.575 0.557
## prnt.empl.blStay at Home Parent -0.010 -0.053 -0.121
## prnt.empl.blUnemployed -0.004 -0.086 -0.120 0.184
## prnt.empl.blOther -0.029 -0.025 -0.029 0.171
## neighb_phenx_avg_p.bl.cm 0.049 0.044 0.049 0.040
## overall.income.bl[>=50K & <100K] -0.273 -0.194 -0.156 -0.022
## overall.income.bl[<50k] -0.420 -0.390 -0.375 -0.051
## overall.income.bl[Don't Know or Refuse] -0.292 -0.283 -0.268 -0.087
## sex.blFemale 0.010 0.014 -0.013 0.017
## reshist_addr1_no2_2016_aavg_bl.c533 0.008 -0.009 -0.033 0.008
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.002 -0.008 0.009 0.015
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.160
## neighb_phenx_avg_p.bl.cm 0.032 -0.014
## overall.income.bl[>=50K & <100K] -0.010 -0.039 0.053
## overall.income.bl[<50k] -0.082 -0.127 0.127 0.550
## overall.income.bl[Don't Know or Refuse] -0.082 -0.112 0.069 0.436
## sex.blFemale 0.038 0.020 0.016 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 -0.005 -0.006 0.083 -0.023
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.011 -0.019 0.010 -0.004
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.616
## sex.blFemale -0.002 -0.003
## reshist_addr1_no2_2016_aavg_bl.c533 -0.033 -0.007 0.004
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.007 0.005 -0.003 0.009
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -1.6591629 -0.3334675 -0.2302656 0.2378951 37.2688755
##
## Number of Observations: 23857
## Number of Groups: 21anova(anxdep_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 337.5860 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 3.3983 0.0653
## interview_age.c9.y 1 14510 22.3540 <.0001
## race_ethnicity.bl 3 9307 32.5402 <.0001
## high.educ.bl 4 9307 5.8412 0.0001
## prnt.empl.bl 3 9307 10.1838 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 85.4505 <.0001
## overall.income.bl 3 9307 7.6765 <.0001
## sex.bl 1 9307 5.8272 0.0158
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 4.3902 0.0362
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 8.0941 0.0044# r2_efron(anxdep_zinb_r)
# anxdep_zinb_r$logLik
#
# r2_efron(anxdep_nb_r$lme)
# anxdep_nb_r$logLikAssumption checking for ZINB Models
#Check outlier/residuals with this df
anxdep_res <- df_cc
anxdep_res$level1_resid.raw <- residuals(anxdep_zinb_r)
anxdep_res$level1_resid.pearson <- residuals(anxdep_zinb_r, type="pearson")
#Add predicted values (Yhat)
anxdep_res$cbcl_scr_syn_anxdep_r_predicted <- predict(anxdep_zinb_r,anxdep_res,type="response")
#Incidence
anxdep_res$incidence <- estimate.probability(anxdep_res$cbcl_scr_syn_anxdep_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(anxdep_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(anxdep_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(anxdep_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
### Residuals vs Y (CBCL Outcome) Plot
plot(anxdep_res$level1_resid.pearson, anxdep_res$cbcl_scr_syn_anxdep_r)
### Residuals vs Yhat Plot
plot(anxdep_res$level1_resid.pearson, anxdep_res$cbcl_scr_syn_anxdep_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(anxdep_res)),anxdep_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(anxdep_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(anxdep_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Withdrawn/Depressed
CBCL + AP Longitudinal Models
withdep_zinb_r <- glmm.zinb(cbcl_scr_syn_withdep_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 15
## Computational time: 2.478 minutessummary(withdep_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1165029
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 1.206047 0.806354
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_withdep_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -0.8001088 0.07060800 14510
## reshist_addr1_pm252016aa_bl.c5 0.0086739 0.01569056 9307
## interview_age.c9.y 0.1231426 0.01308412 14510
## race_ethnicity.blHispanic -0.0081732 0.04636963 9307
## race_ethnicity.blBlack -0.3070256 0.05232687 9307
## race_ethnicity.blOther 0.0204093 0.04718607 9307
## high.educ.blBachelor 0.0768304 0.04009043 9307
## high.educ.blSome College 0.1946432 0.04547991 9307
## high.educ.blHS Diploma/GED 0.0918529 0.06366810 9307
## high.educ.bl< HS Diploma 0.1319487 0.08124908 9307
## prnt.empl.blStay at Home Parent 0.0783155 0.04012623 9307
## prnt.empl.blUnemployed 0.2383777 0.06480874 9307
## prnt.empl.blOther 0.2801519 0.05706675 9307
## neighb_phenx_avg_p.bl.cm -0.1449913 0.01673625 9307
## overall.income.bl[>=50K & <100K] 0.1611700 0.04040551 9307
## overall.income.bl[<50k] 0.2823832 0.05029365 9307
## overall.income.bl[Don't Know or Refuse] 0.2445229 0.06313333 9307
## sex.blFemale -0.0721862 0.02947994 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0036946 0.00369410 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0105938 0.00412280 14510
## t-value p-value
## (Intercept) -11.331702 0.0000
## reshist_addr1_pm252016aa_bl.c5 0.552809 0.5804
## interview_age.c9.y 9.411606 0.0000
## race_ethnicity.blHispanic -0.176261 0.8601
## race_ethnicity.blBlack -5.867456 0.0000
## race_ethnicity.blOther 0.432529 0.6654
## high.educ.blBachelor 1.916428 0.0553
## high.educ.blSome College 4.279761 0.0000
## high.educ.blHS Diploma/GED 1.442683 0.1491
## high.educ.bl< HS Diploma 1.624002 0.1044
## prnt.empl.blStay at Home Parent 1.951728 0.0510
## prnt.empl.blUnemployed 3.678172 0.0002
## prnt.empl.blOther 4.909196 0.0000
## neighb_phenx_avg_p.bl.cm -8.663308 0.0000
## overall.income.bl[>=50K & <100K] 3.988812 0.0001
## overall.income.bl[<50k] 5.614688 0.0000
## overall.income.bl[Don't Know or Refuse] 3.873119 0.0001
## sex.blFemale -2.448655 0.0144
## reshist_addr1_no2_2016_aavg_bl.c533 -1.000139 0.3173
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -2.569567 0.0102
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.418
## interview_age.c9.y -0.337 0.404
## race_ethnicity.blHispanic -0.022 -0.090 0.004
## race_ethnicity.blBlack -0.024 -0.040 0.007
## race_ethnicity.blOther -0.092 -0.034 0.003
## high.educ.blBachelor -0.203 -0.004 -0.001
## high.educ.blSome College -0.141 -0.020 0.003
## high.educ.blHS Diploma/GED -0.084 -0.012 0.006
## high.educ.bl< HS Diploma -0.036 -0.021 -0.001
## prnt.empl.blStay at Home Parent -0.089 -0.019 0.003
## prnt.empl.blUnemployed -0.028 -0.005 0.001
## prnt.empl.blOther -0.043 0.002 0.008
## neighb_phenx_avg_p.bl.cm -0.193 0.063 -0.003
## overall.income.bl[>=50K & <100K] -0.144 -0.014 0.002
## overall.income.bl[<50k] -0.077 -0.028 -0.001
## overall.income.bl[Don't Know or Refuse] -0.071 -0.028 -0.002
## sex.blFemale -0.198 -0.005 0.003
## reshist_addr1_no2_2016_aavg_bl.c533 -0.547 -0.207 0.003
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.290 -0.465 -0.868
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.363
## race_ethnicity.blOther 0.291 0.262
## high.educ.blBachelor -0.019 -0.013 -0.001
## high.educ.blSome College -0.113 -0.089 -0.024 0.464
## high.educ.blHS Diploma/GED -0.145 -0.149 -0.008 0.341
## high.educ.bl< HS Diploma -0.174 -0.076 -0.009 0.273
## prnt.empl.blStay at Home Parent 0.041 0.091 0.017 -0.027
## prnt.empl.blUnemployed 0.007 -0.042 0.009 -0.006
## prnt.empl.blOther 0.042 0.011 -0.009 -0.011
## neighb_phenx_avg_p.bl.cm 0.030 0.140 0.041 -0.006
## overall.income.bl[>=50K & <100K] -0.088 -0.058 -0.008 -0.174
## overall.income.bl[<50k] -0.145 -0.180 -0.078 -0.162
## overall.income.bl[Don't Know or Refuse] -0.103 -0.126 -0.062 -0.102
## sex.blFemale -0.008 -0.015 -0.018 0.017
## reshist_addr1_no2_2016_aavg_bl.c533 -0.055 -0.075 -0.024 0.015
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.002 -0.001 0.000
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.508
## high.educ.bl< HS Diploma 0.421 0.388
## prnt.empl.blStay at Home Parent -0.013 -0.050 -0.094
## prnt.empl.blUnemployed -0.007 -0.068 -0.098 0.152
## prnt.empl.blOther -0.032 -0.008 -0.017 0.162
## neighb_phenx_avg_p.bl.cm 0.061 0.057 0.051 0.027
## overall.income.bl[>=50K & <100K] -0.276 -0.172 -0.116 -0.030
## overall.income.bl[<50k] -0.418 -0.369 -0.312 -0.053
## overall.income.bl[Don't Know or Refuse] -0.255 -0.243 -0.226 -0.077
## sex.blFemale 0.024 0.015 -0.002 -0.005
## reshist_addr1_no2_2016_aavg_bl.c533 0.024 0.007 -0.013 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.002 0.004
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.135
## neighb_phenx_avg_p.bl.cm 0.018 0.005
## overall.income.bl[>=50K & <100K] -0.017 -0.051 0.078
## overall.income.bl[<50k] -0.099 -0.142 0.150 0.513
## overall.income.bl[Don't Know or Refuse] -0.080 -0.099 0.085 0.369
## sex.blFemale 0.018 0.020 0.029 -0.008
## reshist_addr1_no2_2016_aavg_bl.c533 -0.010 -0.003 0.100 -0.007
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 -0.007 0.000 -0.002
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.497
## sex.blFemale -0.006 0.007
## reshist_addr1_no2_2016_aavg_bl.c533 -0.010 0.000 -0.001
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.002 0.000 -0.003
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.7577250 -0.6298805 -0.5045120 0.3858890 4.6880642
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(withdep_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.5499231 0.2649912
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -3.687353 0.13413415 23817
## reshist_addr1_pm252016aa_bl.c5 0.090518 0.01630247 23817
## interview_age.c9.y -0.161258 0.02046086 23817
## race_ethnicity.blHispanic 0.152709 0.03327454 23817
## race_ethnicity.blBlack 0.559740 0.03597775 23817
## race_ethnicity.blOther 0.016267 0.03527193 23817
## high.educ.blBachelor 0.273600 0.02666866 23817
## high.educ.blSome College 0.298743 0.03246994 23817
## high.educ.blHS Diploma/GED 0.205809 0.04895767 23817
## high.educ.bl< HS Diploma 0.642096 0.05816722 23817
## prnt.empl.blStay at Home Parent -0.620143 0.03530450 23817
## prnt.empl.blUnemployed -0.151483 0.05013473 23817
## prnt.empl.blOther -0.042648 0.04319550 23817
## neighb_phenx_avg_p.bl.cm 0.465801 0.01405981 23817
## overall.income.bl[>=50K & <100K] -0.488245 0.02932071 23817
## overall.income.bl[<50k] -0.821929 0.03885078 23817
## overall.income.bl[Don't Know or Refuse] 0.322457 0.03746702 23817
## sex.blFemale 0.205124 0.02085760 23817
## reshist_addr1_no2_2016_aavg_bl.c533 0.003134 0.00310367 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.014003 0.00680183 23817
## t-value p-value
## (Intercept) -27.49004 0.0000
## reshist_addr1_pm252016aa_bl.c5 5.55242 0.0000
## interview_age.c9.y -7.88131 0.0000
## race_ethnicity.blHispanic 4.58937 0.0000
## race_ethnicity.blBlack 15.55794 0.0000
## race_ethnicity.blOther 0.46119 0.6447
## high.educ.blBachelor 10.25924 0.0000
## high.educ.blSome College 9.20062 0.0000
## high.educ.blHS Diploma/GED 4.20382 0.0000
## high.educ.bl< HS Diploma 11.03880 0.0000
## prnt.empl.blStay at Home Parent -17.56557 0.0000
## prnt.empl.blUnemployed -3.02152 0.0025
## prnt.empl.blOther -0.98734 0.3235
## neighb_phenx_avg_p.bl.cm 33.12994 0.0000
## overall.income.bl[>=50K & <100K] -16.65189 0.0000
## overall.income.bl[<50k] -21.15606 0.0000
## overall.income.bl[Don't Know or Refuse] 8.60643 0.0000
## sex.blFemale 9.83448 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 1.00992 0.3125
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -2.05870 0.0395
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.259
## interview_age.c9.y -0.247 0.566
## race_ethnicity.blHispanic -0.019 -0.030 0.008
## race_ethnicity.blBlack -0.012 0.005 0.028
## race_ethnicity.blOther -0.045 0.017 0.020
## high.educ.blBachelor -0.079 0.006 -0.012
## high.educ.blSome College -0.054 -0.001 -0.005
## high.educ.blHS Diploma/GED -0.033 0.018 0.016
## high.educ.bl< HS Diploma -0.005 -0.027 -0.008
## prnt.empl.blStay at Home Parent -0.028 -0.006 -0.001
## prnt.empl.blUnemployed -0.010 -0.003 -0.004
## prnt.empl.blOther -0.016 0.023 0.016
## neighb_phenx_avg_p.bl.cm -0.101 0.035 0.007
## overall.income.bl[>=50K & <100K] -0.030 -0.010 0.007
## overall.income.bl[<50k] -0.010 -0.013 -0.003
## overall.income.bl[Don't Know or Refuse] -0.013 -0.030 -0.001
## sex.blFemale -0.080 -0.014 -0.004
## reshist_addr1_no2_2016_aavg_bl.c533 -0.223 -0.204 -0.001
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.210 -0.660 -0.855
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.352
## race_ethnicity.blOther 0.278 0.230
## high.educ.blBachelor -0.007 -0.002 0.017
## high.educ.blSome College -0.115 -0.108 0.007 0.439
## high.educ.blHS Diploma/GED -0.134 -0.141 0.005 0.300
## high.educ.bl< HS Diploma -0.157 -0.073 0.007 0.258
## prnt.empl.blStay at Home Parent 0.038 0.083 0.018 -0.036
## prnt.empl.blUnemployed 0.013 -0.026 0.017 -0.024
## prnt.empl.blOther 0.036 -0.013 -0.022 -0.027
## neighb_phenx_avg_p.bl.cm 0.014 0.099 0.030 -0.004
## overall.income.bl[>=50K & <100K] -0.102 -0.086 -0.023 -0.140
## overall.income.bl[<50k] -0.122 -0.191 -0.061 -0.114
## overall.income.bl[Don't Know or Refuse] -0.110 -0.168 -0.066 -0.094
## sex.blFemale -0.004 -0.016 -0.009 0.008
## reshist_addr1_no2_2016_aavg_bl.c533 -0.064 -0.116 -0.044 0.001
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.014 -0.017 0.004
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.440
## high.educ.bl< HS Diploma 0.397 0.356
## prnt.empl.blStay at Home Parent -0.020 -0.046 -0.104
## prnt.empl.blUnemployed -0.016 -0.065 -0.128 0.112
## prnt.empl.blOther -0.042 -0.024 -0.045 0.113
## neighb_phenx_avg_p.bl.cm 0.047 0.032 0.040 0.009
## overall.income.bl[>=50K & <100K] -0.273 -0.161 -0.107 -0.018
## overall.income.bl[<50k] -0.362 -0.324 -0.318 -0.028
## overall.income.bl[Don't Know or Refuse] -0.279 -0.271 -0.271 -0.076
## sex.blFemale 0.018 0.007 -0.015 0.002
## reshist_addr1_no2_2016_aavg_bl.c533 0.008 -0.015 -0.034 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.003 -0.010 0.015 0.011
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.115
## neighb_phenx_avg_p.bl.cm 0.029 -0.018
## overall.income.bl[>=50K & <100K] -0.010 -0.049 0.066
## overall.income.bl[<50k] -0.076 -0.114 0.109 0.405
## overall.income.bl[Don't Know or Refuse] -0.079 -0.106 0.069 0.366
## sex.blFemale 0.030 0.010 0.028 -0.018
## reshist_addr1_no2_2016_aavg_bl.c533 -0.010 -0.009 0.112 -0.021
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.008 -0.014 -0.012 -0.005
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.491
## sex.blFemale -0.015 0.001
## reshist_addr1_no2_2016_aavg_bl.c533 -0.033 -0.013 0.008
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 0.001 0.009 0.000
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.5666723 -0.4949188 0.1670057 0.2648371 41.7956765
##
## Number of Observations: 23857
## Number of Groups: 21anova(withdep_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 228.43817 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 2.70550 0.1000
## interview_age.c9.y 1 14510 202.78340 <.0001
## race_ethnicity.bl 3 9307 7.49358 0.0001
## high.educ.bl 4 9307 28.00592 <.0001
## prnt.empl.bl 3 9307 17.99566 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 90.53312 <.0001
## overall.income.bl 3 9307 11.37439 <.0001
## sex.bl 1 9307 5.99550 0.0144
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 1.01734 0.3132
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 6.60268 0.0102Assumption checking for ZINB Models
#Check outlier/residuals with this df
withdep_res <- df_cc
withdep_res$level1_resid.raw <- residuals(withdep_zinb_r)
withdep_res$level1_resid.pearson <- residuals(withdep_zinb_r, type="pearson")
#Add predicted values (Yhat)
withdep_res$cbcl_scr_syn_withdep_r_predicted <- predict(withdep_zinb_r,withdep_res,type="response")
#Incidence
withdep_res$incidence <- estimate.probability(withdep_res$cbcl_scr_syn_withdep_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(withdep_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(withdep_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : pseudoinverse used at 0## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : neighborhood radius 7.4176e-05## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : reciprocal condition number 1.0709e-14## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : There are other near singularities as well. 1.3755e-09## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : pseudoinverse used at 0## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : neighborhood radius
## 7.4176e-05## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : reciprocal condition
## number 1.0709e-14## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : There are other near
## singularities as well. 1.3755e-09
#pm2.5
ggplot(withdep_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : pseudoinverse used at 0## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : neighborhood radius 7.4176e-05## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : reciprocal condition number 1.0709e-14## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : There are other near singularities as well. 1.3755e-09## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : pseudoinverse used at 0## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : neighborhood radius
## 7.4176e-05## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : reciprocal condition
## number 1.0709e-14## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : There are other near
## singularities as well. 1.3755e-09
### Residuals vs Y (CBCL Outcome) Plot
plot(withdep_res$level1_resid.pearson, withdep_res$cbcl_scr_syn_withdep_r)
### Residuals vs Yhat Plot
plot(withdep_res$level1_resid.pearson, withdep_res$cbcl_scr_syn_withdep_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(withdep_res)),withdep_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(withdep_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(withdep_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Attention
CBCL + AP Longitudinal Models
attention_zinb_r <- glmm.zinb(cbcl_scr_syn_attention_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 14
## Computational time: 2.495 minutessummary(attention_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1347084
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 1.139804 0.9078062
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_attention_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) 0.7494678 0.06541076 14510
## reshist_addr1_pm252016aa_bl.c5 -0.0047347 0.01409363 9307
## interview_age.c9.y -0.0283232 0.00948437 14510
## race_ethnicity.blHispanic -0.0494567 0.04150637 9307
## race_ethnicity.blBlack -0.0491935 0.04589533 9307
## race_ethnicity.blOther 0.0655003 0.04196140 9307
## high.educ.blBachelor 0.1536221 0.03529227 9307
## high.educ.blSome College 0.1941088 0.04033107 9307
## high.educ.blHS Diploma/GED 0.0641563 0.05660592 9307
## high.educ.bl< HS Diploma 0.0395986 0.07349492 9307
## prnt.empl.blStay at Home Parent -0.0561871 0.03592320 9307
## prnt.empl.blUnemployed 0.1410724 0.05812388 9307
## prnt.empl.blOther 0.1867426 0.05116620 9307
## neighb_phenx_avg_p.bl.cm -0.1196020 0.01495060 9307
## overall.income.bl[>=50K & <100K] 0.0823427 0.03573809 9307
## overall.income.bl[<50k] 0.1561066 0.04476151 9307
## overall.income.bl[Don't Know or Refuse] 0.0490506 0.05614247 9307
## sex.blFemale -0.4361122 0.02617347 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0063955 0.00344429 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0025515 0.00301549 14510
## t-value p-value
## (Intercept) 11.457868 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.335948 0.7369
## interview_age.c9.y -2.986299 0.0028
## race_ethnicity.blHispanic -1.191544 0.2335
## race_ethnicity.blBlack -1.071863 0.2838
## race_ethnicity.blOther 1.560966 0.1186
## high.educ.blBachelor 4.352855 0.0000
## high.educ.blSome College 4.812886 0.0000
## high.educ.blHS Diploma/GED 1.133385 0.2571
## high.educ.bl< HS Diploma 0.538794 0.5900
## prnt.empl.blStay at Home Parent -1.564090 0.1178
## prnt.empl.blUnemployed 2.427099 0.0152
## prnt.empl.blOther 3.649726 0.0003
## neighb_phenx_avg_p.bl.cm -7.999811 0.0000
## overall.income.bl[>=50K & <100K] 2.304059 0.0212
## overall.income.bl[<50k] 3.487519 0.0005
## overall.income.bl[Don't Know or Refuse] 0.873681 0.3823
## sex.blFemale -16.662379 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -1.856848 0.0634
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.846127 0.3975
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.384
## interview_age.c9.y -0.251 0.315
## race_ethnicity.blHispanic -0.027 -0.077 0.003
## race_ethnicity.blBlack -0.023 -0.029 0.006
## race_ethnicity.blOther -0.088 -0.024 0.004
## high.educ.blBachelor -0.195 0.000 0.000
## high.educ.blSome College -0.132 -0.016 0.002
## high.educ.blHS Diploma/GED -0.078 -0.009 0.004
## high.educ.bl< HS Diploma -0.032 -0.016 -0.002
## prnt.empl.blStay at Home Parent -0.087 -0.017 0.001
## prnt.empl.blUnemployed -0.025 -0.002 0.000
## prnt.empl.blOther -0.040 0.001 0.006
## neighb_phenx_avg_p.bl.cm -0.189 0.057 -0.003
## overall.income.bl[>=50K & <100K] -0.131 -0.016 0.000
## overall.income.bl[<50k] -0.066 -0.027 -0.001
## overall.income.bl[Don't Know or Refuse] -0.064 -0.028 -0.002
## sex.blFemale -0.185 -0.003 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 -0.535 -0.234 0.003
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.218 -0.361 -0.873
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.356
## race_ethnicity.blOther 0.287 0.263
## high.educ.blBachelor -0.017 -0.010 0.000
## high.educ.blSome College -0.106 -0.079 -0.021 0.460
## high.educ.blHS Diploma/GED -0.139 -0.143 -0.006 0.338
## high.educ.bl< HS Diploma -0.164 -0.072 -0.009 0.265
## prnt.empl.blStay at Home Parent 0.044 0.091 0.016 -0.026
## prnt.empl.blUnemployed 0.010 -0.041 0.010 -0.010
## prnt.empl.blOther 0.041 0.009 -0.014 -0.011
## neighb_phenx_avg_p.bl.cm 0.028 0.138 0.041 -0.003
## overall.income.bl[>=50K & <100K] -0.089 -0.061 -0.011 -0.173
## overall.income.bl[<50k] -0.143 -0.184 -0.080 -0.161
## overall.income.bl[Don't Know or Refuse] -0.097 -0.126 -0.057 -0.101
## sex.blFemale -0.006 -0.017 -0.016 0.014
## reshist_addr1_no2_2016_aavg_bl.c533 -0.059 -0.084 -0.032 0.013
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 -0.002 -0.001
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.503
## high.educ.bl< HS Diploma 0.410 0.379
## prnt.empl.blStay at Home Parent -0.013 -0.050 -0.092
## prnt.empl.blUnemployed -0.010 -0.069 -0.099 0.148
## prnt.empl.blOther -0.033 -0.013 -0.020 0.158
## neighb_phenx_avg_p.bl.cm 0.062 0.057 0.053 0.026
## overall.income.bl[>=50K & <100K] -0.278 -0.175 -0.114 -0.032
## overall.income.bl[<50k] -0.421 -0.369 -0.310 -0.054
## overall.income.bl[Don't Know or Refuse] -0.254 -0.241 -0.219 -0.078
## sex.blFemale 0.021 0.014 -0.003 -0.005
## reshist_addr1_no2_2016_aavg_bl.c533 0.019 0.005 -0.018 0.008
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.003 0.003
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.133
## neighb_phenx_avg_p.bl.cm 0.019 0.002
## overall.income.bl[>=50K & <100K] -0.016 -0.050 0.082
## overall.income.bl[<50k] -0.101 -0.138 0.150 0.508
## overall.income.bl[Don't Know or Refuse] -0.079 -0.099 0.084 0.363
## sex.blFemale 0.019 0.020 0.029 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.012 -0.001 0.101 -0.007
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.005 0.001 -0.001
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.488
## sex.blFemale -0.008 0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.015 -0.001 0.000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.001 -0.001 -0.003
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.5119456 -0.7122067 -0.2814738 0.4195745 3.7883930
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(attention_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.5614799 0.399084
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -5.515229 0.18061096 23817
## reshist_addr1_pm252016aa_bl.c5 0.123391 0.03327092 23817
## interview_age.c9.y 0.208024 0.04337886 23817
## race_ethnicity.blHispanic 0.277392 0.07348386 23817
## race_ethnicity.blBlack 0.855752 0.07050106 23817
## race_ethnicity.blOther 0.510489 0.07158396 23817
## high.educ.blBachelor 0.063779 0.06643831 23817
## high.educ.blSome College 0.332285 0.07321077 23817
## high.educ.blHS Diploma/GED 1.024946 0.08689514 23817
## high.educ.bl< HS Diploma 2.022126 0.09136890 23817
## prnt.empl.blStay at Home Parent 0.150021 0.05547594 23817
## prnt.empl.blUnemployed -0.363908 0.09168686 23817
## prnt.empl.blOther -0.162201 0.08711050 23817
## neighb_phenx_avg_p.bl.cm 0.452065 0.02565896 23817
## overall.income.bl[>=50K & <100K] -0.361035 0.06823295 23817
## overall.income.bl[<50k] -0.046953 0.07611816 23817
## overall.income.bl[Don't Know or Refuse] 0.090411 0.08660639 23817
## sex.blFemale 0.745166 0.04423475 23817
## reshist_addr1_no2_2016_aavg_bl.c533 -0.032491 0.00600455 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.020282 0.01281459 23817
## t-value p-value
## (Intercept) -30.536514 0.0000
## reshist_addr1_pm252016aa_bl.c5 3.708666 0.0002
## interview_age.c9.y 4.795515 0.0000
## race_ethnicity.blHispanic 3.774872 0.0002
## race_ethnicity.blBlack 12.138149 0.0000
## race_ethnicity.blOther 7.131331 0.0000
## high.educ.blBachelor 0.959979 0.3371
## high.educ.blSome College 4.538746 0.0000
## high.educ.blHS Diploma/GED 11.795203 0.0000
## high.educ.bl< HS Diploma 22.131446 0.0000
## prnt.empl.blStay at Home Parent 2.704250 0.0069
## prnt.empl.blUnemployed -3.969028 0.0001
## prnt.empl.blOther -1.862015 0.0626
## neighb_phenx_avg_p.bl.cm 17.618217 0.0000
## overall.income.bl[>=50K & <100K] -5.291206 0.0000
## overall.income.bl[<50k] -0.616842 0.5373
## overall.income.bl[Don't Know or Refuse] 1.043933 0.2965
## sex.blFemale 16.845716 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -5.411105 0.0000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -1.582700 0.1135
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.449
## interview_age.c9.y -0.469 0.625
## race_ethnicity.blHispanic -0.052 -0.032 0.011
## race_ethnicity.blBlack -0.051 0.012 0.041
## race_ethnicity.blOther -0.092 0.017 0.018
## high.educ.blBachelor -0.139 -0.002 -0.018
## high.educ.blSome College -0.104 -0.016 0.001
## high.educ.blHS Diploma/GED -0.085 0.013 0.014
## high.educ.bl< HS Diploma -0.052 -0.004 0.000
## prnt.empl.blStay at Home Parent -0.074 -0.008 0.004
## prnt.empl.blUnemployed -0.029 0.013 0.016
## prnt.empl.blOther -0.028 0.022 0.018
## neighb_phenx_avg_p.bl.cm -0.133 0.026 0.004
## overall.income.bl[>=50K & <100K] -0.066 -0.002 0.009
## overall.income.bl[<50k] -0.027 -0.019 -0.011
## overall.income.bl[Don't Know or Refuse] -0.030 -0.032 0.003
## sex.blFemale -0.158 -0.015 0.005
## reshist_addr1_no2_2016_aavg_bl.c533 -0.317 -0.175 0.015
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.399 -0.710 -0.873
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.477
## race_ethnicity.blOther 0.381 0.375
## high.educ.blBachelor -0.006 0.007 0.019
## high.educ.blSome College -0.097 -0.076 -0.006 0.485
## high.educ.blHS Diploma/GED -0.151 -0.143 -0.001 0.417
## high.educ.bl< HS Diploma -0.182 -0.103 -0.002 0.400
## prnt.empl.blStay at Home Parent 0.047 0.095 0.035 -0.030
## prnt.empl.blUnemployed 0.018 -0.029 -0.003 -0.018
## prnt.empl.blOther 0.048 -0.007 -0.017 -0.018
## neighb_phenx_avg_p.bl.cm 0.015 0.115 0.027 -0.007
## overall.income.bl[>=50K & <100K] -0.088 -0.098 -0.043 -0.156
## overall.income.bl[<50k] -0.157 -0.215 -0.099 -0.154
## overall.income.bl[Don't Know or Refuse] -0.129 -0.162 -0.086 -0.119
## sex.blFemale 0.000 -0.017 -0.013 0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.067 -0.124 -0.063 0.011
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.004 -0.029 -0.014 0.016
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.616
## high.educ.bl< HS Diploma 0.609 0.637
## prnt.empl.blStay at Home Parent -0.014 -0.047 -0.109
## prnt.empl.blUnemployed -0.005 -0.059 -0.110 0.182
## prnt.empl.blOther -0.017 -0.005 -0.045 0.171
## neighb_phenx_avg_p.bl.cm 0.029 0.033 0.064 0.045
## overall.income.bl[>=50K & <100K] -0.310 -0.242 -0.209 -0.019
## overall.income.bl[<50k] -0.433 -0.450 -0.460 -0.043
## overall.income.bl[Don't Know or Refuse] -0.324 -0.357 -0.375 -0.089
## sex.blFemale 0.008 0.008 -0.001 0.015
## reshist_addr1_no2_2016_aavg_bl.c533 0.024 -0.007 -0.048 0.024
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.005 -0.003 0.008 0.008
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.138
## neighb_phenx_avg_p.bl.cm 0.033 -0.015
## overall.income.bl[>=50K & <100K] -0.008 -0.041 0.048
## overall.income.bl[<50k] -0.051 -0.115 0.118 0.528
## overall.income.bl[Don't Know or Refuse] -0.055 -0.097 0.068 0.433
## sex.blFemale 0.038 0.012 0.016 -0.017
## reshist_addr1_no2_2016_aavg_bl.c533 -0.010 -0.016 0.090 -0.015
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.014 -0.014 -0.004 -0.006
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.651
## sex.blFemale -0.008 -0.017
## reshist_addr1_no2_2016_aavg_bl.c533 -0.045 -0.011 0.008
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.014 0.009 0.005 -0.010
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -1.6609515 -0.2735201 -0.1785196 0.1605542 40.7544065
##
## Number of Observations: 23857
## Number of Groups: 21anova(attention_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 305.99064 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 0.23777 0.6258
## interview_age.c9.y 1 14510 58.60303 <.0001
## race_ethnicity.bl 3 9307 4.67899 0.0029
## high.educ.bl 4 9307 21.82688 <.0001
## prnt.empl.bl 3 9307 11.50626 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 63.43170 <.0001
## overall.income.bl 3 9307 4.06511 0.0068
## sex.bl 1 9307 277.68187 <.0001
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 3.45771 0.0630
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 0.71593 0.3975Assumption checking for ZINB Models
#Check outlier/residuals with this df
attention_res <- df_cc
attention_res$level1_resid.raw <- residuals(attention_zinb_r)
attention_res$level1_resid.pearson <- residuals(attention_zinb_r, type="pearson")
#Add predicted values (Yhat)
attention_res$cbcl_scr_syn_attention_r_predicted <- predict(attention_zinb_r,attention_res,type="response")
#Incidence
attention_res$incidence <- estimate.probability(attention_res$cbcl_scr_syn_attention_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(attention_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(attention_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(attention_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
### Residuals vs Y (CBCL Outcome) Plot
plot(attention_res$level1_resid.pearson, attention_res$cbcl_scr_syn_attention_r)
### Residuals vs Yhat Plot
plot(attention_res$level1_resid.pearson, attention_res$cbcl_scr_syn_attention_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(attention_res)),attention_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(attention_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(attention_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Rulebreak
CBCL + AP Longitudinal Models
rulebreak_zinb_r <- glmm.zinb(cbcl_scr_syn_rulebreak_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 15
## Computational time: 2.39 minutessummary(rulebreak_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1319985
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 1.193222 0.7910848
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_rulebreak_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -0.4608865 0.07201545 14510
## reshist_addr1_pm252016aa_bl.c5 -0.0086035 0.01574688 9307
## interview_age.c9.y -0.0111185 0.01307453 14510
## race_ethnicity.blHispanic -0.0625569 0.04606489 9307
## race_ethnicity.blBlack 0.1011862 0.05032717 9307
## race_ethnicity.blOther 0.0854988 0.04677108 9307
## high.educ.blBachelor 0.1800712 0.03982702 9307
## high.educ.blSome College 0.3288204 0.04485979 9307
## high.educ.blHS Diploma/GED 0.1893917 0.06237381 9307
## high.educ.bl< HS Diploma 0.2308833 0.08012045 9307
## prnt.empl.blStay at Home Parent -0.0324430 0.04008399 9307
## prnt.empl.blUnemployed 0.1794433 0.06335055 9307
## prnt.empl.blOther 0.2562250 0.05599409 9307
## neighb_phenx_avg_p.bl.cm -0.1236551 0.01651205 9307
## overall.income.bl[>=50K & <100K] 0.1495291 0.04016566 9307
## overall.income.bl[<50k] 0.3314525 0.04959594 9307
## overall.income.bl[Don't Know or Refuse] 0.2610857 0.06199701 9307
## sex.blFemale -0.4244973 0.02922905 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0055482 0.00376578 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0073172 0.00406477 14510
## t-value p-value
## (Intercept) -6.399828 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.546362 0.5848
## interview_age.c9.y -0.850393 0.3951
## race_ethnicity.blHispanic -1.358017 0.1745
## race_ethnicity.blBlack 2.010569 0.0444
## race_ethnicity.blOther 1.828028 0.0676
## high.educ.blBachelor 4.521333 0.0000
## high.educ.blSome College 7.329958 0.0000
## high.educ.blHS Diploma/GED 3.036398 0.0024
## high.educ.bl< HS Diploma 2.881703 0.0040
## prnt.empl.blStay at Home Parent -0.809376 0.4183
## prnt.empl.blUnemployed 2.832545 0.0046
## prnt.empl.blOther 4.575929 0.0000
## neighb_phenx_avg_p.bl.cm -7.488779 0.0000
## overall.income.bl[>=50K & <100K] 3.722810 0.0002
## overall.income.bl[<50k] 6.683056 0.0000
## overall.income.bl[Don't Know or Refuse] 4.211262 0.0000
## sex.blFemale -14.523131 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -1.473311 0.1407
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -1.800149 0.0719
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.410
## interview_age.c9.y -0.315 0.381
## race_ethnicity.blHispanic -0.020 -0.083 0.004
## race_ethnicity.blBlack -0.022 -0.037 0.008
## race_ethnicity.blOther -0.089 -0.031 0.004
## high.educ.blBachelor -0.202 0.000 0.000
## high.educ.blSome College -0.143 -0.016 0.002
## high.educ.blHS Diploma/GED -0.086 -0.008 0.005
## high.educ.bl< HS Diploma -0.038 -0.016 -0.002
## prnt.empl.blStay at Home Parent -0.084 -0.018 0.002
## prnt.empl.blUnemployed -0.025 -0.003 0.000
## prnt.empl.blOther -0.040 0.002 0.008
## neighb_phenx_avg_p.bl.cm -0.189 0.058 -0.003
## overall.income.bl[>=50K & <100K] -0.140 -0.014 0.001
## overall.income.bl[<50k] -0.075 -0.027 -0.001
## overall.income.bl[Don't Know or Refuse] -0.071 -0.027 -0.002
## sex.blFemale -0.181 -0.006 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 -0.541 -0.213 0.003
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.273 -0.436 -0.873
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.367
## race_ethnicity.blOther 0.293 0.273
## high.educ.blBachelor -0.020 -0.012 0.001
## high.educ.blSome College -0.106 -0.080 -0.021 0.473
## high.educ.blHS Diploma/GED -0.143 -0.146 -0.008 0.351
## high.educ.bl< HS Diploma -0.167 -0.076 -0.011 0.278
## prnt.empl.blStay at Home Parent 0.043 0.093 0.016 -0.030
## prnt.empl.blUnemployed 0.012 -0.038 0.010 -0.008
## prnt.empl.blOther 0.047 0.016 -0.011 -0.013
## neighb_phenx_avg_p.bl.cm 0.034 0.140 0.046 -0.003
## overall.income.bl[>=50K & <100K] -0.091 -0.066 -0.015 -0.171
## overall.income.bl[<50k] -0.146 -0.186 -0.080 -0.161
## overall.income.bl[Don't Know or Refuse] -0.101 -0.128 -0.057 -0.101
## sex.blFemale -0.008 -0.019 -0.016 0.011
## reshist_addr1_no2_2016_aavg_bl.c533 -0.063 -0.084 -0.029 0.012
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.002 -0.002 -0.002
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.516
## high.educ.bl< HS Diploma 0.423 0.389
## prnt.empl.blStay at Home Parent -0.016 -0.051 -0.099
## prnt.empl.blUnemployed -0.008 -0.068 -0.100 0.152
## prnt.empl.blOther -0.034 -0.016 -0.023 0.162
## neighb_phenx_avg_p.bl.cm 0.061 0.054 0.050 0.030
## overall.income.bl[>=50K & <100K] -0.275 -0.175 -0.116 -0.028
## overall.income.bl[<50k] -0.420 -0.367 -0.310 -0.050
## overall.income.bl[Don't Know or Refuse] -0.258 -0.244 -0.222 -0.070
## sex.blFemale 0.022 0.014 -0.003 -0.007
## reshist_addr1_no2_2016_aavg_bl.c533 0.022 0.007 -0.015 0.002
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.003 0.004
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.138
## neighb_phenx_avg_p.bl.cm 0.023 0.003
## overall.income.bl[>=50K & <100K] -0.015 -0.049 0.079
## overall.income.bl[<50k] -0.102 -0.140 0.151 0.519
## overall.income.bl[Don't Know or Refuse] -0.082 -0.101 0.084 0.375
## sex.blFemale 0.014 0.017 0.031 -0.010
## reshist_addr1_no2_2016_aavg_bl.c533 -0.013 -0.004 0.099 -0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.007 0.001 -0.002
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.503
## sex.blFemale -0.010 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 -0.011 0.000 0.002
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.002 0.000 -0.003
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.9548528 -0.6410680 -0.4932774 0.3944843 3.9742633
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(rulebreak_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.4646952 0.2627375
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -4.858859 0.12108106 23817
## reshist_addr1_pm252016aa_bl.c5 -0.030596 0.01800879 23817
## interview_age.c9.y 0.233989 0.01942862 23817
## race_ethnicity.blHispanic 0.189782 0.03463825 23817
## race_ethnicity.blBlack -0.218182 0.04808695 23817
## race_ethnicity.blOther -0.036684 0.03540483 23817
## high.educ.blBachelor -0.254884 0.02764554 23817
## high.educ.blSome College -0.638935 0.03767951 23817
## high.educ.blHS Diploma/GED -0.130831 0.04971140 23817
## high.educ.bl< HS Diploma 0.014054 0.05841238 23817
## prnt.empl.blStay at Home Parent -0.004982 0.02979314 23817
## prnt.empl.blUnemployed -0.163256 0.05885508 23817
## prnt.empl.blOther -0.072334 0.04961089 23817
## neighb_phenx_avg_p.bl.cm 0.387341 0.01464331 23817
## overall.income.bl[>=50K & <100K] -0.277322 0.03047146 23817
## overall.income.bl[<50k] -0.018113 0.03917084 23817
## overall.income.bl[Don't Know or Refuse] 0.410501 0.04187221 23817
## sex.blFemale 1.210516 0.02444207 23817
## reshist_addr1_no2_2016_aavg_bl.c533 0.014326 0.00329089 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.009678 0.00644513 23817
## t-value p-value
## (Intercept) -40.12898 0.0000
## reshist_addr1_pm252016aa_bl.c5 -1.69894 0.0893
## interview_age.c9.y 12.04351 0.0000
## race_ethnicity.blHispanic 5.47897 0.0000
## race_ethnicity.blBlack -4.53725 0.0000
## race_ethnicity.blOther -1.03613 0.3002
## high.educ.blBachelor -9.21973 0.0000
## high.educ.blSome College -16.95710 0.0000
## high.educ.blHS Diploma/GED -2.63180 0.0085
## high.educ.bl< HS Diploma 0.24059 0.8099
## prnt.empl.blStay at Home Parent -0.16723 0.8672
## prnt.empl.blUnemployed -2.77386 0.0055
## prnt.empl.blOther -1.45804 0.1448
## neighb_phenx_avg_p.bl.cm 26.45174 0.0000
## overall.income.bl[>=50K & <100K] -9.10102 0.0000
## overall.income.bl[<50k] -0.46240 0.6438
## overall.income.bl[Don't Know or Refuse] 9.80366 0.0000
## sex.blFemale 49.52589 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 4.35330 0.0000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1.50160 0.1332
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.304
## interview_age.c9.y -0.326 0.593
## race_ethnicity.blHispanic -0.024 -0.032 0.011
## race_ethnicity.blBlack -0.019 0.006 0.024
## race_ethnicity.blOther -0.049 0.009 0.009
## high.educ.blBachelor -0.066 -0.004 -0.016
## high.educ.blSome College -0.035 -0.012 -0.006
## high.educ.blHS Diploma/GED -0.017 -0.005 0.006
## high.educ.bl< HS Diploma 0.013 -0.047 -0.018
## prnt.empl.blStay at Home Parent -0.042 -0.020 -0.002
## prnt.empl.blUnemployed -0.013 -0.010 -0.005
## prnt.empl.blOther -0.017 0.017 0.011
## neighb_phenx_avg_p.bl.cm -0.121 0.033 0.006
## overall.income.bl[>=50K & <100K] -0.054 0.003 0.004
## overall.income.bl[<50k] -0.024 -0.007 0.001
## overall.income.bl[Don't Know or Refuse] -0.030 -0.015 0.005
## sex.blFemale -0.153 -0.007 0.009
## reshist_addr1_no2_2016_aavg_bl.c533 -0.270 -0.204 0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.271 -0.718 -0.833
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.285
## race_ethnicity.blOther 0.289 0.180
## high.educ.blBachelor -0.004 -0.007 0.009
## high.educ.blSome College -0.128 -0.086 -0.002 0.342
## high.educ.blHS Diploma/GED -0.183 -0.134 0.001 0.270
## high.educ.bl< HS Diploma -0.199 -0.080 0.010 0.237
## prnt.empl.blStay at Home Parent 0.048 0.074 0.012 -0.042
## prnt.empl.blUnemployed 0.001 -0.038 0.004 -0.017
## prnt.empl.blOther 0.030 -0.013 -0.025 -0.030
## neighb_phenx_avg_p.bl.cm 0.017 0.084 0.020 -0.013
## overall.income.bl[>=50K & <100K] -0.089 -0.057 -0.016 -0.143
## overall.income.bl[<50k] -0.128 -0.144 -0.065 -0.145
## overall.income.bl[Don't Know or Refuse] -0.094 -0.096 -0.067 -0.094
## sex.blFemale -0.003 -0.012 -0.015 0.005
## reshist_addr1_no2_2016_aavg_bl.c533 -0.048 -0.077 -0.024 0.013
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.003 -0.013 -0.009 0.010
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.405
## high.educ.bl< HS Diploma 0.379 0.403
## prnt.empl.blStay at Home Parent -0.030 -0.075 -0.129
## prnt.empl.blUnemployed -0.015 -0.056 -0.097 0.132
## prnt.empl.blOther -0.026 0.005 -0.029 0.137
## neighb_phenx_avg_p.bl.cm 0.039 0.039 0.053 0.025
## overall.income.bl[>=50K & <100K] -0.220 -0.134 -0.091 -0.014
## overall.income.bl[<50k] -0.382 -0.392 -0.355 -0.018
## overall.income.bl[Don't Know or Refuse] -0.227 -0.264 -0.241 -0.082
## sex.blFemale 0.004 -0.003 -0.012 0.001
## reshist_addr1_no2_2016_aavg_bl.c533 0.017 -0.007 -0.026 0.017
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 0.002 0.025 0.015
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.099
## neighb_phenx_avg_p.bl.cm 0.027 0.001
## overall.income.bl[>=50K & <100K] -0.009 -0.051 0.089
## overall.income.bl[<50k] -0.069 -0.132 0.113 0.383
## overall.income.bl[Don't Know or Refuse] -0.059 -0.099 0.072 0.301
## sex.blFemale 0.026 0.010 0.034 -0.014
## reshist_addr1_no2_2016_aavg_bl.c533 -0.006 -0.014 0.124 0.000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.010 -0.011 -0.006 0.000
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.456
## sex.blFemale -0.004 0.016
## reshist_addr1_no2_2016_aavg_bl.c533 -0.026 -0.004 0.010
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.002 0.002 -0.009
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.0220695 -0.4361089 0.1495185 0.2581541 45.5351413
##
## Number of Observations: 23857
## Number of Groups: 21anova(rulebreak_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 175.30571 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 1.45330 0.2280
## interview_age.c9.y 1 14510 26.62996 <.0001
## race_ethnicity.bl 3 9307 30.05610 <.0001
## high.educ.bl 4 9307 50.15661 <.0001
## prnt.empl.bl 3 9307 16.73595 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 64.61592 <.0001
## overall.income.bl 3 9307 14.60484 <.0001
## sex.bl 1 9307 210.84597 <.0001
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 2.18877 0.1391
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 3.24054 0.0719Assumption checking for ZINB Models
#Check outlier/residuals with this df
rulebreak_res <- df_cc
rulebreak_res$level1_resid.raw <- residuals(rulebreak_zinb_r)
rulebreak_res$level1_resid.pearson <- residuals(rulebreak_zinb_r, type="pearson")
#Add predicted values (Yhat)
rulebreak_res$cbcl_scr_syn_rulebreak_r_predicted <- predict(rulebreak_zinb_r,rulebreak_res,type="response")
#Incidence
rulebreak_res$incidence <- estimate.probability(rulebreak_res$cbcl_scr_syn_rulebreak_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(rulebreak_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(rulebreak_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : pseudoinverse used at 0## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : neighborhood radius 7.303e-05## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : reciprocal condition number 1.3261e-14## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : There are other near singularities as well. 1.3333e-09## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : pseudoinverse used at 0## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : neighborhood radius
## 7.303e-05## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : reciprocal condition
## number 1.3261e-14## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : There are other near
## singularities as well. 1.3333e-09
#pm2.5
ggplot(rulebreak_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : pseudoinverse used at 0## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : neighborhood radius 7.303e-05## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : reciprocal condition number 1.3261e-14## Warning in simpleLoess(y, x, w, span, degree = degree, parametric = parametric,
## : There are other near singularities as well. 1.3333e-09## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : pseudoinverse used at 0## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : neighborhood radius
## 7.303e-05## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : reciprocal condition
## number 1.3261e-14## Warning in predLoess(object$y, object$x, newx = if (is.null(newdata)) object$x
## else if (is.data.frame(newdata))
## as.matrix(model.frame(delete.response(terms(object)), : There are other near
## singularities as well. 1.3333e-09
### Residuals vs Y (CBCL Outcome) Plot
plot(rulebreak_res$level1_resid.pearson, rulebreak_res$cbcl_scr_syn_rulebreak_r)
### Residuals vs Yhat Plot
plot(rulebreak_res$level1_resid.pearson, rulebreak_res$cbcl_scr_syn_rulebreak_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(rulebreak_res)),rulebreak_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(rulebreak_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(rulebreak_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Aggressive
CBCL + AP Longitudinal Models
aggressive_zinb_r <- glmm.zinb(cbcl_scr_syn_aggressive_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid,
zi_fixed = ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, zi_random = ~1|abcd_site, data = df_cc)## Computational iterations: 11
## Computational time: 1.733 minutessummary(aggressive_zinb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1353382
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 1.145622 0.9763728
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_aggressive_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) 0.7125267 0.06613630 14510
## reshist_addr1_pm252016aa_bl.c5 -0.0079460 0.01430213 9307
## interview_age.c9.y -0.0289695 0.00994498 14510
## race_ethnicity.blHispanic -0.0506505 0.04198298 9307
## race_ethnicity.blBlack -0.1868707 0.04674669 9307
## race_ethnicity.blOther -0.0837799 0.04281735 9307
## high.educ.blBachelor 0.0872491 0.03579956 9307
## high.educ.blSome College 0.1510548 0.04084542 9307
## high.educ.blHS Diploma/GED 0.0261277 0.05734511 9307
## high.educ.bl< HS Diploma 0.0932123 0.07377417 9307
## prnt.empl.blStay at Home Parent 0.0250469 0.03624081 9307
## prnt.empl.blUnemployed 0.2143845 0.05878427 9307
## prnt.empl.blOther 0.1827674 0.05189617 9307
## neighb_phenx_avg_p.bl.cm -0.1278335 0.01512435 9307
## overall.income.bl[>=50K & <100K] 0.1229781 0.03618114 9307
## overall.income.bl[<50k] 0.2496497 0.04526850 9307
## overall.income.bl[Don't Know or Refuse] 0.1385538 0.05688246 9307
## sex.blFemale -0.2494005 0.02647986 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0049382 0.00348323 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0055511 0.00315186 14510
## t-value p-value
## (Intercept) 10.773610 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.555580 0.5785
## interview_age.c9.y -2.912975 0.0036
## race_ethnicity.blHispanic -1.206454 0.2277
## race_ethnicity.blBlack -3.997518 0.0001
## race_ethnicity.blOther -1.956681 0.0504
## high.educ.blBachelor 2.437155 0.0148
## high.educ.blSome College 3.698206 0.0002
## high.educ.blHS Diploma/GED 0.455623 0.6487
## high.educ.bl< HS Diploma 1.263481 0.2064
## prnt.empl.blStay at Home Parent 0.691125 0.4895
## prnt.empl.blUnemployed 3.646970 0.0003
## prnt.empl.blOther 3.521790 0.0004
## neighb_phenx_avg_p.bl.cm -8.452168 0.0000
## overall.income.bl[>=50K & <100K] 3.398957 0.0007
## overall.income.bl[<50k] 5.514867 0.0000
## overall.income.bl[Don't Know or Refuse] 2.435792 0.0149
## sex.blFemale -9.418498 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -1.417718 0.1563
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -1.761213 0.0782
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.385
## interview_age.c9.y -0.260 0.322
## race_ethnicity.blHispanic -0.024 -0.078 0.004
## race_ethnicity.blBlack -0.022 -0.029 0.006
## race_ethnicity.blOther -0.086 -0.023 0.003
## high.educ.blBachelor -0.195 0.002 0.000
## high.educ.blSome College -0.132 -0.016 0.002
## high.educ.blHS Diploma/GED -0.080 -0.008 0.004
## high.educ.bl< HS Diploma -0.032 -0.016 -0.001
## prnt.empl.blStay at Home Parent -0.085 -0.018 0.002
## prnt.empl.blUnemployed -0.026 -0.003 0.000
## prnt.empl.blOther -0.039 0.001 0.006
## neighb_phenx_avg_p.bl.cm -0.186 0.058 -0.002
## overall.income.bl[>=50K & <100K] -0.131 -0.016 0.000
## overall.income.bl[<50k] -0.066 -0.029 -0.001
## overall.income.bl[Don't Know or Refuse] -0.063 -0.030 -0.002
## sex.blFemale -0.187 -0.003 0.003
## reshist_addr1_no2_2016_aavg_bl.c533 -0.533 -0.236 0.003
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.224 -0.370 -0.869
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.355
## race_ethnicity.blOther 0.287 0.261
## high.educ.blBachelor -0.022 -0.017 -0.003
## high.educ.blSome College -0.112 -0.086 -0.026 0.461
## high.educ.blHS Diploma/GED -0.144 -0.148 -0.010 0.339
## high.educ.bl< HS Diploma -0.169 -0.078 -0.013 0.268
## prnt.empl.blStay at Home Parent 0.043 0.092 0.018 -0.031
## prnt.empl.blUnemployed 0.010 -0.041 0.011 -0.010
## prnt.empl.blOther 0.040 0.011 -0.011 -0.014
## neighb_phenx_avg_p.bl.cm 0.029 0.137 0.041 -0.005
## overall.income.bl[>=50K & <100K] -0.087 -0.060 -0.010 -0.176
## overall.income.bl[<50k] -0.141 -0.180 -0.079 -0.161
## overall.income.bl[Don't Know or Refuse] -0.095 -0.123 -0.055 -0.101
## sex.blFemale -0.008 -0.018 -0.019 0.014
## reshist_addr1_no2_2016_aavg_bl.c533 -0.059 -0.083 -0.032 0.015
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.001 -0.002 -0.001
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.502
## high.educ.bl< HS Diploma 0.413 0.381
## prnt.empl.blStay at Home Parent -0.016 -0.050 -0.095
## prnt.empl.blUnemployed -0.011 -0.069 -0.099 0.149
## prnt.empl.blOther -0.033 -0.014 -0.020 0.159
## neighb_phenx_avg_p.bl.cm 0.061 0.056 0.049 0.028
## overall.income.bl[>=50K & <100K] -0.276 -0.174 -0.116 -0.028
## overall.income.bl[<50k] -0.417 -0.367 -0.311 -0.052
## overall.income.bl[Don't Know or Refuse] -0.253 -0.240 -0.220 -0.075
## sex.blFemale 0.021 0.014 -0.005 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 0.021 0.008 -0.016 0.006
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.002 0.003
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.134
## neighb_phenx_avg_p.bl.cm 0.022 0.004
## overall.income.bl[>=50K & <100K] -0.014 -0.048 0.080
## overall.income.bl[<50k] -0.100 -0.138 0.151 0.508
## overall.income.bl[Don't Know or Refuse] -0.077 -0.097 0.083 0.363
## sex.blFemale 0.019 0.017 0.027 -0.005
## reshist_addr1_no2_2016_aavg_bl.c533 -0.011 -0.002 0.098 -0.010
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 -0.006 0.000 -0.001
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.488
## sex.blFemale -0.006 0.009
## reshist_addr1_no2_2016_aavg_bl.c533 -0.017 -0.004 -0.001
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.002 0.000 -0.003
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -3.4152997 -0.7126800 -0.3414064 0.4036700 4.0640651
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345summary(aggressive_zinb_r$zi.fit)## Linear mixed-effects model fit by maximum likelihood
## Data: data
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept) Residual
## StdDev: 0.2872063 0.4680178
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: zp ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) -4.562223 0.12721169 23817
## reshist_addr1_pm252016aa_bl.c5 0.111760 0.02877556 23817
## interview_age.c9.y 0.311412 0.03539220 23817
## race_ethnicity.blHispanic 0.355667 0.05967951 23817
## race_ethnicity.blBlack 0.726116 0.06154681 23817
## race_ethnicity.blOther 0.464283 0.05665542 23817
## high.educ.blBachelor 0.145535 0.04969539 23817
## high.educ.blSome College 0.229520 0.05834128 23817
## high.educ.blHS Diploma/GED 0.389079 0.08006616 23817
## high.educ.bl< HS Diploma 1.018741 0.08726699 23817
## prnt.empl.blStay at Home Parent -0.034355 0.05173919 23817
## prnt.empl.blUnemployed 0.231090 0.07356334 23817
## prnt.empl.blOther -0.160481 0.07944159 23817
## neighb_phenx_avg_p.bl.cm 0.265650 0.02231635 23817
## overall.income.bl[>=50K & <100K] -0.296345 0.05187634 23817
## overall.income.bl[<50k] -0.538582 0.06532682 23817
## overall.income.bl[Don't Know or Refuse] 0.038402 0.07166424 23817
## sex.blFemale 0.202557 0.03645698 23817
## reshist_addr1_no2_2016_aavg_bl.c533 -0.004932 0.00516650 23817
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.024467 0.01086372 23817
## t-value p-value
## (Intercept) -35.86324 0.0000
## reshist_addr1_pm252016aa_bl.c5 3.88385 0.0001
## interview_age.c9.y 8.79890 0.0000
## race_ethnicity.blHispanic 5.95962 0.0000
## race_ethnicity.blBlack 11.79779 0.0000
## race_ethnicity.blOther 8.19486 0.0000
## high.educ.blBachelor 2.92855 0.0034
## high.educ.blSome College 3.93409 0.0001
## high.educ.blHS Diploma/GED 4.85947 0.0000
## high.educ.bl< HS Diploma 11.67385 0.0000
## prnt.empl.blStay at Home Parent -0.66400 0.5067
## prnt.empl.blUnemployed 3.14137 0.0017
## prnt.empl.blOther -2.02012 0.0434
## neighb_phenx_avg_p.bl.cm 11.90385 0.0000
## overall.income.bl[>=50K & <100K] -5.71252 0.0000
## overall.income.bl[<50k] -8.24443 0.0000
## overall.income.bl[Don't Know or Refuse] 0.53586 0.5921
## sex.blFemale 5.55606 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -0.95465 0.3398
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -2.25217 0.0243
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.532
## interview_age.c9.y -0.570 0.645
## race_ethnicity.blHispanic -0.048 -0.040 0.007
## race_ethnicity.blBlack -0.045 -0.002 0.032
## race_ethnicity.blOther -0.098 0.004 0.012
## high.educ.blBachelor -0.153 -0.001 -0.007
## high.educ.blSome College -0.108 -0.008 0.003
## high.educ.blHS Diploma/GED -0.075 0.007 0.017
## high.educ.bl< HS Diploma -0.028 -0.030 -0.008
## prnt.empl.blStay at Home Parent -0.075 -0.008 0.005
## prnt.empl.blUnemployed -0.031 0.002 0.006
## prnt.empl.blOther -0.032 0.015 0.018
## neighb_phenx_avg_p.bl.cm -0.167 0.036 0.005
## overall.income.bl[>=50K & <100K] -0.072 -0.009 -0.001
## overall.income.bl[<50k] -0.025 -0.022 -0.011
## overall.income.bl[Don't Know or Refuse] -0.031 -0.036 -0.007
## sex.blFemale -0.157 -0.004 0.009
## reshist_addr1_no2_2016_aavg_bl.c533 -0.407 -0.172 0.007
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.481 -0.741 -0.860
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.408
## race_ethnicity.blOther 0.340 0.308
## high.educ.blBachelor -0.006 -0.003 0.018
## high.educ.blSome College -0.114 -0.093 -0.005 0.452
## high.educ.blHS Diploma/GED -0.142 -0.145 0.005 0.340
## high.educ.bl< HS Diploma -0.169 -0.090 0.002 0.317
## prnt.empl.blStay at Home Parent 0.048 0.100 0.021 -0.029
## prnt.empl.blUnemployed 0.018 -0.027 0.014 -0.016
## prnt.empl.blOther 0.042 0.003 -0.019 -0.020
## neighb_phenx_avg_p.bl.cm 0.016 0.122 0.027 0.002
## overall.income.bl[>=50K & <100K] -0.098 -0.087 -0.022 -0.154
## overall.income.bl[<50k] -0.133 -0.200 -0.075 -0.138
## overall.income.bl[Don't Know or Refuse] -0.107 -0.145 -0.061 -0.105
## sex.blFemale -0.003 -0.023 -0.011 0.011
## reshist_addr1_no2_2016_aavg_bl.c533 -0.064 -0.105 -0.044 0.014
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.018 -0.011 0.006
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.507
## high.educ.bl< HS Diploma 0.490 0.476
## prnt.empl.blStay at Home Parent -0.017 -0.051 -0.113
## prnt.empl.blUnemployed -0.009 -0.079 -0.129 0.176
## prnt.empl.blOther -0.029 -0.011 -0.030 0.144
## neighb_phenx_avg_p.bl.cm 0.055 0.048 0.062 0.029
## overall.income.bl[>=50K & <100K] -0.291 -0.188 -0.148 -0.022
## overall.income.bl[<50k] -0.396 -0.379 -0.387 -0.047
## overall.income.bl[Don't Know or Refuse] -0.291 -0.293 -0.314 -0.090
## sex.blFemale 0.017 0.011 -0.005 0.004
## reshist_addr1_no2_2016_aavg_bl.c533 0.020 -0.001 -0.020 0.014
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 -0.006 0.019 0.006
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.140
## neighb_phenx_avg_p.bl.cm 0.033 -0.007
## overall.income.bl[>=50K & <100K] -0.017 -0.046 0.073
## overall.income.bl[<50k] -0.097 -0.124 0.129 0.474
## overall.income.bl[Don't Know or Refuse] -0.100 -0.111 0.089 0.391
## sex.blFemale 0.036 0.013 0.023 0.001
## reshist_addr1_no2_2016_aavg_bl.c533 -0.009 -0.008 0.107 -0.017
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.015 -0.005 0.002
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.556
## sex.blFemale 0.000 0.001
## reshist_addr1_no2_2016_aavg_bl.c533 -0.027 -0.003 0.005
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.011 0.016 -0.001 -0.007
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -1.1649791 -0.3653236 -0.2726505 0.2194252 19.4199661
##
## Number of Observations: 23857
## Number of Groups: 21anova(aggressive_zinb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 341.9834 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 0.1523 0.6964
## interview_age.c9.y 1 14510 82.2311 <.0001
## race_ethnicity.bl 3 9307 2.4685 0.0601
## high.educ.bl 4 9307 21.8524 <.0001
## prnt.empl.bl 3 9307 13.1186 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 81.7904 <.0001
## overall.income.bl 3 9307 10.0399 <.0001
## sex.bl 1 9307 88.7303 <.0001
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 2.0241 0.1549
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 3.1019 0.0782Assumption checking for ZINB Models
#Check outlier/residuals with this df
aggressive_res <- df_cc
aggressive_res$level1_resid.raw <- residuals(aggressive_zinb_r)
aggressive_res$level1_resid.pearson <- residuals(aggressive_zinb_r, type="pearson")
#Add predicted values (Yhat)
aggressive_res$cbcl_scr_syn_aggressive_r_predicted <- predict(aggressive_zinb_r,aggressive_res,type="response")
#Incidence
aggressive_res$incidence <- estimate.probability(aggressive_res$cbcl_scr_syn_aggressive_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using ZINB, so make sure to check below plots
hist(aggressive_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(aggressive_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(aggressive_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
### Residuals vs Y (CBCL Outcome) Plot
plot(aggressive_res$level1_resid.pearson, aggressive_res$cbcl_scr_syn_aggressive_r)
### Residuals vs Yhat Plot
plot(aggressive_res$level1_resid.pearson, aggressive_res$cbcl_scr_syn_aggressive_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(aggressive_res)),aggressive_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(aggressive_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(aggressive_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
Total Problems
CBCL + AP Longitudinal Models
Convergence Error with ZINB Model - thinking it’s b/c total problems might not be heavily zero inflated
hist(df_cc$cbcl_scr_syn_totprob_r)
hist(df_cc$cbcl_scr_syn_totprob_r[df_cc$eventname=="Baseline"])
hist(df_cc$cbcl_scr_syn_totprob_r[df_cc$eventname=="1-year"])
hist(df_cc$cbcl_scr_syn_totprob_r[df_cc$eventname=="2-year"])
# totprob_zinb_r <- glmm.zinb(cbcl_scr_syn_totprob_r ~ reshist_addr1_pm252016aa_bl*interview_age + race_ethnicity.1 + high.educ_bl+ prnt.empl.alltp + neighb_phenx_avg_p + overall.income.alltp + sex, random = ~1|abcd_site/subjectid,
# zi_fixed = ~ reshist_addr1_pm252016aa_bl*interview_age + race_ethnicity.1 + high.educ_bl+ prnt.empl.alltp + neighb_phenx_avg_p + overall.income.alltp + sex, zi_random = ~1|abcd_site/subjectid, data = df_cc)
#
# summary(totprob_zinb_r)
# summary(totprob_zinb_r$zi.fit)
# anova(totprob_zinb_r)
#Trying just normal negative binomial model due to distribution of outcome
totprob_nb_r <- glmm.nb(cbcl_scr_syn_totprob_r ~ reshist_addr1_pm252016aa_bl.c5*interview_age.c9.y + race_ethnicity.bl + high.educ.bl+ prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533, random = ~1|abcd_site/subjectid, data = df_cc)## Computational iterations: 6
## Computational time: 0.481 minutessummary(totprob_nb_r)## Linear mixed-effects model fit by maximum likelihood
## Data: df_cc
## AIC BIC logLik
## NA NA NA
##
## Random effects:
## Formula: ~1 | abcd_site
## (Intercept)
## StdDev: 0.1280292
##
## Formula: ~1 | subjectid %in% abcd_site
## (Intercept) Residual
## StdDev: 0.8686449 1.212098
##
## Variance function:
## Structure: fixed weights
## Formula: ~invwt
## Fixed effects: cbcl_scr_syn_totprob_r ~ reshist_addr1_pm252016aa_bl.c5 * interview_age.c9.y + race_ethnicity.bl + high.educ.bl + prnt.empl.bl + neighb_phenx_avg_p.bl.cm + overall.income.bl + sex.bl + reshist_addr1_no2_2016_aavg_bl.c533
## Value Std.Error DF
## (Intercept) 2.6451853 0.05268093 14510
## reshist_addr1_pm252016aa_bl.c5 -0.0004004 0.01105319 9307
## interview_age.c9.y -0.0209051 0.00798467 14510
## race_ethnicity.blHispanic -0.0565100 0.03097169 9307
## race_ethnicity.blBlack -0.2392913 0.03433506 9307
## race_ethnicity.blOther -0.0416651 0.03118715 9307
## high.educ.blBachelor 0.0876493 0.02603209 9307
## high.educ.blSome College 0.1263014 0.02987890 9307
## high.educ.blHS Diploma/GED -0.0542696 0.04209925 9307
## high.educ.bl< HS Diploma -0.0595857 0.05431965 9307
## prnt.empl.blStay at Home Parent -0.0033125 0.02651104 9307
## prnt.empl.blUnemployed 0.1377946 0.04353720 9307
## prnt.empl.blOther 0.1929010 0.03828081 9307
## neighb_phenx_avg_p.bl.cm -0.1317072 0.01112065 9307
## overall.income.bl[>=50K & <100K] 0.0964629 0.02639029 9307
## overall.income.bl[<50k] 0.1784080 0.03324592 9307
## overall.income.bl[Don't Know or Refuse] 0.0479272 0.04161909 9307
## sex.blFemale -0.1875365 0.01933200 9307
## reshist_addr1_no2_2016_aavg_bl.c533 -0.0051972 0.00265276 9307
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.0081013 0.00255056 14510
## t-value p-value
## (Intercept) 50.21144 0.0000
## reshist_addr1_pm252016aa_bl.c5 -0.03623 0.9711
## interview_age.c9.y -2.61815 0.0088
## race_ethnicity.blHispanic -1.82457 0.0681
## race_ethnicity.blBlack -6.96930 0.0000
## race_ethnicity.blOther -1.33597 0.1816
## high.educ.blBachelor 3.36697 0.0008
## high.educ.blSome College 4.22711 0.0000
## high.educ.blHS Diploma/GED -1.28909 0.1974
## high.educ.bl< HS Diploma -1.09695 0.2727
## prnt.empl.blStay at Home Parent -0.12495 0.9006
## prnt.empl.blUnemployed 3.16498 0.0016
## prnt.empl.blOther 5.03910 0.0000
## neighb_phenx_avg_p.bl.cm -11.84348 0.0000
## overall.income.bl[>=50K & <100K] 3.65524 0.0003
## overall.income.bl[<50k] 5.36631 0.0000
## overall.income.bl[Don't Know or Refuse] 1.15157 0.2495
## sex.blFemale -9.70083 0.0000
## reshist_addr1_no2_2016_aavg_bl.c533 -1.95918 0.0501
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -3.17629 0.0015
## Correlation:
## (Intr) rs_1_252016_.5 in_.9.
## reshist_addr1_pm252016aa_bl.c5 -0.374
## interview_age.c9.y -0.265 0.339
## race_ethnicity.blHispanic -0.026 -0.066 0.004
## race_ethnicity.blBlack -0.019 -0.020 0.007
## race_ethnicity.blOther -0.082 -0.015 0.004
## high.educ.blBachelor -0.178 0.002 0.000
## high.educ.blSome College -0.120 -0.012 0.003
## high.educ.blHS Diploma/GED -0.072 -0.006 0.005
## high.educ.bl< HS Diploma -0.027 -0.015 -0.002
## prnt.empl.blStay at Home Parent -0.081 -0.015 0.001
## prnt.empl.blUnemployed -0.025 -0.001 0.000
## prnt.empl.blOther -0.038 0.003 0.007
## neighb_phenx_avg_p.bl.cm -0.174 0.047 -0.003
## overall.income.bl[>=50K & <100K] -0.116 -0.016 0.001
## overall.income.bl[<50k] -0.055 -0.025 -0.001
## overall.income.bl[Don't Know or Refuse] -0.055 -0.027 -0.002
## sex.blFemale -0.174 -0.004 0.003
## reshist_addr1_no2_2016_aavg_bl.c533 -0.500 -0.247 0.004
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.228 -0.392 -0.866
## rc_t.H rc_t.B rc_t.O hgh..B
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack 0.351
## race_ethnicity.blOther 0.287 0.259
## high.educ.blBachelor -0.019 -0.013 0.000
## high.educ.blSome College -0.112 -0.085 -0.024 0.455
## high.educ.blHS Diploma/GED -0.144 -0.145 -0.006 0.333
## high.educ.bl< HS Diploma -0.165 -0.074 -0.009 0.263
## prnt.empl.blStay at Home Parent 0.043 0.091 0.018 -0.025
## prnt.empl.blUnemployed 0.011 -0.040 0.010 -0.009
## prnt.empl.blOther 0.040 0.009 -0.012 -0.013
## neighb_phenx_avg_p.bl.cm 0.028 0.134 0.039 -0.003
## overall.income.bl[>=50K & <100K] -0.091 -0.063 -0.015 -0.173
## overall.income.bl[<50k] -0.143 -0.184 -0.081 -0.159
## overall.income.bl[Don't Know or Refuse] -0.095 -0.126 -0.060 -0.100
## sex.blFemale -0.008 -0.018 -0.017 0.014
## reshist_addr1_no2_2016_aavg_bl.c533 -0.058 -0.089 -0.034 0.014
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.000 -0.001 -0.002 -0.001
## hg..SC h..HSD h..<HD p..aHP
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED 0.497
## high.educ.bl< HS Diploma 0.408 0.377
## prnt.empl.blStay at Home Parent -0.012 -0.048 -0.093
## prnt.empl.blUnemployed -0.009 -0.068 -0.096 0.146
## prnt.empl.blOther -0.033 -0.012 -0.020 0.157
## neighb_phenx_avg_p.bl.cm 0.061 0.055 0.050 0.027
## overall.income.bl[>=50K & <100K] -0.276 -0.172 -0.114 -0.032
## overall.income.bl[<50k] -0.416 -0.367 -0.309 -0.054
## overall.income.bl[Don't Know or Refuse] -0.251 -0.240 -0.221 -0.077
## sex.blFemale 0.022 0.016 -0.004 -0.005
## reshist_addr1_no2_2016_aavg_bl.c533 0.019 0.004 -0.019 0.007
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y -0.001 -0.001 0.003 0.004
## prn..U prn..O n___.. o..[&<
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther 0.130
## neighb_phenx_avg_p.bl.cm 0.022 0.003
## overall.income.bl[>=50K & <100K] -0.015 -0.048 0.081
## overall.income.bl[<50k] -0.100 -0.137 0.150 0.504
## overall.income.bl[Don't Know or Refuse] -0.077 -0.097 0.083 0.360
## sex.blFemale 0.020 0.019 0.026 -0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.010 -0.002 0.102 -0.010
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.002 -0.006 0.001 -0.001
## o..[<5 o..KoR sx.blF r_1_2_
## reshist_addr1_pm252016aa_bl.c5
## interview_age.c9.y
## race_ethnicity.blHispanic
## race_ethnicity.blBlack
## race_ethnicity.blOther
## high.educ.blBachelor
## high.educ.blSome College
## high.educ.blHS Diploma/GED
## high.educ.bl< HS Diploma
## prnt.empl.blStay at Home Parent
## prnt.empl.blUnemployed
## prnt.empl.blOther
## neighb_phenx_avg_p.bl.cm
## overall.income.bl[>=50K & <100K]
## overall.income.bl[<50k]
## overall.income.bl[Don't Know or Refuse] 0.485
## sex.blFemale -0.008 0.006
## reshist_addr1_no2_2016_aavg_bl.c533 -0.022 -0.005 0.000
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 0.001 0.002 0.000 -0.004
##
## Standardized Within-Group Residuals:
## Min Q1 Med Q3 Max
## -2.5189022 -0.6176406 -0.0744057 0.4578181 4.5302533
##
## Number of Observations: 23857
## Number of Groups:
## abcd_site subjectid %in% abcd_site
## 21 9345anova(totprob_nb_r)## numDF denDF F-value p-value
## (Intercept) 1 14510 7133.859 <.0001
## reshist_addr1_pm252016aa_bl.c5 1 9307 0.136 0.7121
## interview_age.c9.y 1 14510 115.715 <.0001
## race_ethnicity.bl 3 9307 2.918 0.0328
## high.educ.bl 4 9307 25.732 <.0001
## prnt.empl.bl 3 9307 17.297 <.0001
## neighb_phenx_avg_p.bl.cm 1 9307 153.599 <.0001
## overall.income.bl 3 9307 10.571 <.0001
## sex.bl 1 9307 94.135 <.0001
## reshist_addr1_no2_2016_aavg_bl.c533 1 9307 3.887 0.0487
## reshist_addr1_pm252016aa_bl.c5:interview_age.c9.y 1 14510 10.089 0.0015Assumption checking for ZINB Models
#Check outlier/residuals with this df
totprob_res <- df_cc
totprob_res$level1_resid.raw <- residuals(totprob_nb_r)
totprob_res$level1_resid.pearson <- residuals(totprob_nb_r, type="pearson")
#Add predicted values (Yhat)
totprob_res$cbcl_scr_syn_totprob_r_predicted <- predict(totprob_nb_r,totprob_res,type="response")
#Incidence
totprob_res$incidence <- estimate.probability(totprob_res$cbcl_scr_syn_totprob_r, method="empirical")
#Plotting histogram of residuals, but may be skewed since using nb, so make sure to check below plots
hist(totprob_res$level1_resid.pearson)
### Incidence vs. X’s Plots
#age
ggplot(totprob_res,aes(incidence,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(totprob_res,aes(incidence,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
### Residuals vs Y (CBCL Outcome) Plot
plot(totprob_res$level1_resid.pearson, totprob_res$cbcl_scr_syn_totprob_r)
### Residuals vs Yhat Plot
plot(totprob_res$level1_resid.pearson, totprob_res$cbcl_scr_syn_totprob_r_predicted)
### Residuals vs Row Plot
plot(as.numeric(rownames(totprob_res)),totprob_res$level1_resid.pearson)
### Residuals vs X’s Plots
#age
ggplot(totprob_res,aes(level1_resid.pearson,interview_age)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'
#pm2.5
ggplot(totprob_res,aes(level1_resid.pearson,reshist_addr1_pm252016aa_bl)) + geom_point(color = "black") + geom_smooth(method = "loess")## `geom_smooth()` using formula = 'y ~ x'